#If you write the algorithm and train it on data that you obtained by paying and/or receiving consent

[i-like-turkey]

AI relies on stolen material. Using it is unethical. And using it write fics is really fucked up even if AO3 allows it.

[dmitrilyalikov]

Why do America’s generations keep getting dumber?

America is the global symbol of individual liberty and opportunity. Defined by capitalism and democracy, the very concepts that have made the U.S. the hallmark in innovative thinking and societal development. With arguably the best ‘system’ in the world able to work at great scale, American renegades have been the frontrunners in many aspects of society many countries wish they could compete with. Walt Disney, Bill Gates, Steve Jobs, Mark Zuckerberg, all American icons for creative thinking and execution. Creative, intelligent men that any company would love to have on their team if they could convince them to come. They’ve accomplished things that some would believe to be impossible, and not only that, they all dropped out of college. The education system failed them. 

The current American educational system was first introduced in the 1910′s during the industrial era to create a scaled up version of a youth knowledge assembly line. Children are crammed into large classrooms and are taught general knowledge to enter the next level of education. The strict regimen of be quiet, listen, and regurgitate what you have heard onto a standardized exam to get a letter grade has been used for over a century. This practice is nowhere near teaching a child to think and solve problems. Tests do not work. They do not represent any more than words on a paper. Example, the Chinese Box Experiment. In short, a Chinese professor inserts a test of different Mandarin characters that a robot on the other side of the door must answer. The robot identified every character correctly and returned the paper. The Professor says “Wow, this pupil understands Mandarin very well!”. She is unaware the answers came from a machine programmed by humans. The robot does not actually understand what is going on, it is simply responding with what it’s been told to do. Understanding is using memory to create predictions. However, this is exactly how school teaches children in America. They program children to respond to an input with a correct output, and those that compute such information correctly, are deemed the brightest. If we are programming children to act as robots, robots will win every time, bar none. The only way to fundamentally beat a robot is to be more human. Humans have creativity, emotional intelligence, morals, historical and societal awareness. Schools are essentially building kids like robots in an assembly line. They are writing code in our brains on how to think, act, and behave in many situations. The smartest natural child can be nurtured in such an environment to become average. 

The most beautiful aspect of a child is its sense of curiosity and creativity. Left to its own, many will fantasize about spaceships and rockets and trains. They will dance on couches, spill their parent’s coffee on the rug, They ask naive questions about complex issues. I was lucky enough as a child as my father would make me understand how any toy or tool worked when I used it. I was made to inquire about the world around me. How does a car engine work? What could make it better? Why do planes not fall from the sky? I was then sent to day school and would be told to shut up and listen to the teacher, because he is smarter than you. What does it mean to be smart then? To know more information and algorithms downloaded into the hippocampus? Memory is not intelligence. Intelligence and consciousness are manifested in the neocortex. The part of the brain that operates high level thought. Children in American society are suppressed and told to remember things to graduate. After a certain point of indoctrinated thinking, children lose their sense of curiosity and are more focused on execution then the process of learning and solving the problem itself. The most commonly asked question in American schools is “Will this be on the test next week?”.

So how can we make this better? This epidemic starts on the very system of education itself. The end goal of school is to obtain a degree, a rough representation of what college taught you, or maybe you were just wily enough to cheat (which is highly incentivized in the ends justify the means environment.). School’s are not obligated to innovate. Colleges are businesses. They force 18 year old children to take on 200 thousand dollar debt decisions. They don't need all that money. The books that cost hundreds of dollars for students, cost 6$ to make. NO INDUSTRY IN THE WORLD HAS A MONOPOLY THAT BOASTS SUCH GREAT PROFIT MARGINS. Colleges have young generations on a string with the rhetoric that a degree is worth such money. Millions of kids cry joyfully over getting into a school, just to give them money that is taken from loans to enslave them once they get out with a degree. College is enslavement. It is a monopolistic business. It is a shame to see such an important factor to human development being exploited for profit. They pay zero taxes on the profits they make. They teach general knowledge in a lecture style. Is that worth it? Why do kids want this? Why do parents make them do this? Because they did it when they were kids? We are in a new age. 

Fast forward over a century later, the digital age. Children have smartphones, smartphones with all the information they need. Why sit in a room listening to someone lecture when you can just look something up? Children are put in classrooms that are part of a school, that are part of a district, that is part of a school board. These scaled up versions of education pump out millions of children with a broad range of general knowledge, or at least that is the intent. Now most of these kids go to college, work a 9-5 job, and start a family and the cycle goes on with their children. That is not fulfillment, that is not happiness for most. The average school tuition has increased by more than 200% while the average salary of college graduates has plateaued since the start of mass schooling. We live in an era of economies of “unscale”. With artificial intelligence and cloud computing, vertically integrated corporations with huge factories and inventory cannot compete with lean, agile startups that rent cloud storage on Amazon Web Services, outsource manufacturing to Chinese factories, and utilize open source Machine Learning algorithms instead of spending great capital to build it all individually. This gives power to creative, niche startups that can effectively run a business from their basement. Think back to the 1990′s. The internet had just gone mainstream, thousands of employees quit their jobs to create internet companies during the Dot-Com Boom before it crashed. They would plan their IPO before even incorporating, this new technology was a home run in their eyes. How does this relate to education? The rapid evolution of technology can be attributed to new platforms. Telecommunications created a global platform for information to be spread from Boston to Australia in an instant, the internet has revolutionized virtually every industry. My generation is growing up in the advent of the AI and cloud computing platform. Essentially, the innovation of big tech platforms should equate to radically different education. However, because school systems have no incentive to change and make less profit, they are still preparing kids for an industrial era to be interchangeable pieces working for large corporations rather than agile startups and small to medium companies. 

Artificial Intelligence will radically change education. Harvard, Stanford, and a few other large brand schools have noticed this trend and created online courses already that use machine learning engines to tailor a course to a students understanding. AI can use big data to understand how a pupil learns, what he/she is struggling in, and create a report on their level of thought that is a perfect representation on what they can do, rather than a vague degree. Many companies such as Microsoft and Google are receptive to this and an increasing number of developers enter the software field with no degrees. Because there is no system that could exemplify a student’s intelligence in the past, an expensive degree was the next best thing and college became a booming business but quite an enslaving process for the children utilizing it. AI can guide a student while virtual classrooms and teachers can connect to children across the globe for real organic conversation. Now, the physical classroom is very important for social development and should still be used to an extent. Perhaps we Americans should look towards Finland, the country with the best ranked educational system in the world. Their primary and secondary schools are incredibly different. School days are 3 hours long, there is no homework, and there are no private schools. The philosophy is that kids should be emancipated from the institutions and be left to be kids and develop intuitiveness organically through real world social experiences. There are no private schools so that rich families send their kids to public schools and those parents make sure the school is up to par with what they can afford.This forces schools nationwide to keep a standard that is universal, much unlike the U.S. with many inner city public schools without internet while capitalistic private and public district schools spend money on football field renovations. 

To create a more productive generation of students, we must “unscale” education, remove private schools, reduce length of school hours, ban or at least regulate student loan firms, set a price ceiling on all college tuitions and utilize the platform of Artificial Intelligence to create a market of one for all students starting from Kindergarten to beyond college. Hiring more teachers and building would effectively make the problem worse. Teachers can be the greatest minds on the planet, but under such a restrictive there is little hope to save a whole generation. Khan Academy has implemented an unscaled online system, leading the way for more personalized education programs. There is little chance this can happen unless this is derived from the Federal Government, which is famously bureaucratic and slow to act especially with education. Changes are needed. This will make children more excited to learn, ask questions and solve the great global issues that are long overdue to be solved. Kids will strengthen critical thinking skills and experience freedom of thought that will create a wave of further technological development and accelerate American education to new heights. 

[schultz88hamilton-blog]

Search engine optimisation - How to Increase Your Rank for Google Search Engine Benefits

Google is at the foremost amongst all lookup engines, that a big volume of web end users are using to search for pertinent websites that Google returns for a certain keyword or keyphrase becoming used. So if you have to actually make your site very popular, that appeals to a huge quantity of prospective clients, then its critical you have to comply with particular set up Search engine optimization methods. Key phrases are the Key to Success of Your Site You have to find those search phrases that are relevant to the variety of concept or subject that you select for to promote products and companies from your website. You have to locate keywords that finds a very good many inquiry via the Google search box but not several internet sites are there that actually offers with that keyword. Now if you scan via these internet sites that are exhibited on the initial page of a Google Research for that keyword, you may possibly locate that there is not ample articles that throws mild on the search term or phrase employed. Listed here you can exploit, by producing inroads. Extremely Educational Articles Prosperous Sites Begin composing wealthy content that are highly educational and beneficial to the type of website visitors who are clicking their way into your site. Additionally keep your web site clean by writing a lot more contents, 2 times a week. Google loves articles that are fresh, scorching and crispy. Not stale kinds, churning in articles or blog posts that most internet sites have already dealt with in excess of and over once again, and standard site visitors have learnt them by coronary heart and does not need it any more. In truth Google is worried, that a excellent numerous high quality visitors are obtaining diverted from Google and as an alternative preferring other lookup engines above Google this sort of as Bing, to search for web sites that has prosperous content and very diverse from the crushed route. In a natural way the reduction of high quality site visitors will imply significantly less of profits earning for Google. That is why Google is hungry for fresh contents from related web sites, catering to a distinct marketplace segment or group. Satisfy Google's Hunger For Clean Content material That is why it is critical that just by catering to the specifications of a distinct market section or goal group won't suffice., You also acquired to search from Google's position of view way too. So write or outsource refreshing articles or blog posts, that are abundant in appropriate data and also in context to the key phrase, primarily based on which you want to obtain foothold as an set up web internet site to reckon with. The Relevance of Back again Back links for Your Site for Large Google Rating Now you want again back links too for your website. as for each the PageRank (Webpage stands for the name of Google co founder Larry Webpage) algorithm, back links are quite critical if a internet site needs to climb the greasy pole to Google look for end result page. A lot more inbound links a site have the much better it is. However quality one way links too count. In other words and phrases if a very established website has a website link in one of its webpages that points to your website then, the web site, as for every Google, has the deserves to occupy the first couple of slots of the first page for a offered research key phrase/phrase. Report Writing Article composing is one particular way of making your site popular, as creating in established article directories, support you to garner more back links that level to your web site. premium backlinks to boost rank are of program market discussion board internet sites, where you can lead often with your useful tips in the dialogue threads relating to your subject of experience and area of interest. That way you can send a lot more back again back links to your site, and thus deliver much more quality specific site visitors from the forum too. Blogs For much more again hyperlinks you can maintain blogs that too are material wealthy and has a back again url to your websites. This way you can increase the number of back back links. Of training course these are all hard work considering that you have to see that the website is routinely updated with wealthy search term dependent related material. Cease Making use of Flash and Graphics in Your Site Chorus from squandering your cherished time and cash in sprucing up your website with flash and graphics considering that the codes are not at all crawled by Google Spiders. So no position in adding individuals codes that Google barely pays interest to. Quit Your Website from Being Sluggish and Sluggish Google does not like individuals internet sites that are gradual and irritating to open rapidly. So do not use as well a lot of video clips and graphics that weighs down your internet site, and makes it open quickly. The Significance of HTML / XML Website Map before Publishing to Google and other Research Engines Yet another thing I desire to draw your attention is about the relevance of a web site map ideally XML website map. It is preferable even so to have each HTML and XML in the site root folder. The explanation is apparent given that diverse search engines have their personal preference. It is therefore quite critical that you have a web site map for each web sites prior to you post to Google. google booster can create a website map for free this sort of as http://www.auditmypc.com/totally free-sitemap-generator.asp. After you have developed your web site map you could then post it to Google. You will have a fair concept if you visit the website map of EzineArticles, which is ezinearticles.com/sitemap.html Stop Plagiarism by Duplicating Content from other Sites Plagiarism is illegal and is an offence for violation of copyright and mental property laws. Google also hates sites that thrives on duplicate content. So ensure that your site has material which is composed in first and no other site has the identical content word to phrase and sentence right after sentence. Material have to be of at least 250 terms Make certain your post is at the very least 250 words in every single webpage of your web site. Also your key phrases need to have to be set in the 1st 250 terms, since Google crawler scans only the first 250 words and phrases and on that foundation it checks the relevancy of the webpage for look for consequence. Keyword Density - Limit it to 2% - three% of whole content material. Use Only 2 to three keywords only Do not use too a lot of the very same search phrases more than and in excess of once more. It would be considered offensive by most lookup engines. The best is to limit it to three% of the general content material. Nevertheless it is prudent that you maintain it to simply 2 to 3 diverse key phrases by itself. Market Sites That has quite Minor Competition but has a Very good Need from Authentic Guests Do not switch your web site into some kind of a tremendous shopping mall, that sells from mouse to elephant all under 1 roof. Locate your niche that caters to a good section of the market place, lets say Area flipping. See that your area of interest is not previously crowded with as well many gamers in the industry. Consider these that has not also several gamers and you can be 1 of them, to combat you way to the top in the Google Lookup Motor rankings, by catering to only a selected clientele. Hope the details offered to you listed here in this write-up will support you to get commenced for a productive job in web advertising and marketing and good results from websites that you make them rewarding employing Search engine optimisation techniques. This is how you can boost the ranking of your web site by Google Research Engines.

[fusiononrketing]

Fusion one news summary from March 16th to March 31st

Hi everyone, we have back with new stuff and knowledge to share with you all. We come up every week to share new information that can help business owners to improve their online business and achieve new goals.

 March 17, 2022

 Google Ad trends that will put you on top

                                                                             In this blog, we will talk about new Google Ad capabilities. There are some new trends to pay attention to.

 Automation: AI helps to analyze how well your ads are performing across different platforms and then offer recommendations on optimizing your ad better. It also helps save time and identifies other issues with your ad. It also helps to automate your bidding and target your audience and predict through rates and conversions. 

 New Local Campaign formats: Local ad campaigns help local businesses to attract more potential customers. Google has not added three new ad formats to promote local businesses better. The new formats include auto-suggest ads, navigation ads, and similar places ads. 

 Google Shopping: Google has implemented updates to accommodate the increased eCommerce demands. Google has also integrated Shopify into its Google Ad platform. Their shopping graph will now begin to pull data (prices, videos, product information from across 

 First-Party data and Google ad: 3rd-party cookies, which drive a lot of online ads, will soon be obsolete. Google announced it would block them on Chrome in 2023. As part of the plan to evolve to “cookie-less browsing,” Google is moving into 1st party data. 1st party data is information a company collects directly from the customers it owns whereas 3rd party data is any data that you want to obtain but is not your own.

 How to use 1st party data when creating your Google ad?

 Google recommends creating segments. You can define your audience segments based on how users interact with your site and what content they view. Google will also automatically add some segments for you based on your audience sources. In doing this, you will add your data segments to ad groups and ad campaigns. 

 Focus more on quality vs. cost: The higher-performing ads are the ones that are generating customers and revenues, not just leads. To make this a success, marketers should be looking closely at how well their ad lines up with their ideal customers. A great way to do this is to train the algorithms within the platform to understand what type of conversion results in a new customer. One way to do this is to integrate your CRM with Google Ads. If that’s not an option, you can pass your data findings back and forth manually. Click on https://fusiononemarketing.com/google-ad-trends-that-will-put-you-on-top/ to find more. 

March 21, 2022

 Evergreen Content: What it is and how to create it in

Evergreen content is content that stays relevant throughout time. We will learn about different types of evergreen content and more in this blog. 

 What is not evergreen content?

Most content is relevant for a short period. It’s led by current news, trends, and other time-sensitive articles that speak to a brief moment in time. Although this type of content will gain a spike in attention soon after being published, interest declines quickly. 

 What is evergreen content?

Evergreen content is search-optimized content that people will always be interested in. It’s informative and not time-sensitive. 

 Evergreen content can be in a variety of formats, some examples that perform well are:

How-To’s

Guides and FAQs

Listicles

Tips

Instructional videos

Publishing content in the above formats doesn’t automatically guarantee you’ve got an evergreen piece. However, these types of content tend to provide real value and longevity to the end-user.

 How to create evergreen content?

 Some ways to create evergreen content include:

-        Find the right topics

-        Proper keyword research

-        Write for beginners

-        Update your evergreen content regularly

 You can learn more on this topic by clicking here https://fusiononemarketing.com/evergreen-content/

March 28, 2022

 Title Tags- How to beat Google title rewrites

 Some factors can reduce your chances of search engines rewriting your title tags.

Why does Google rewrite your title tags?

 The best way to determine what search result to click is by finding the most relevant titles that match the search query. This is the same why Google works hard to provide the best titles that properly describe the contents of the page. 

 Factors for title rewrites

 The factors that can help to keep your title tags the same are:

Too long to too short title length: The title length is an important factor. Any title less than 20 characters had a 50% chance of a Google rewrite, and any title over 60 characters was rewritten 76% of the time. Aiming for title tags between 51-and 60 seems to be the sweet spot for Google.

Match the Title tag to the H1: Pages where the title and the H1 match typically see less rewriting. When the title is different than the H1, Google may choose to use the H1 instead so match your

 Title to your H1 across the board.

Title Separators: The separator that gets the least amount of rewrites, and thus seems to be Google’s favorite is the dash “-“. Your best bet to avoid a rewrite is to use the dash as your title separator.

 Avoid Brackets: Use the parenthesis in your page title tags to avoid rewrite.

Use the right keywords: To help avoid the rewrite, check the top keywords users search to find a specific page and make sure you include it in your title. Click here for more https://fusiononemarketing.com/title-tags/

 March 31, 2022

 Website optimization: How to improve your online rankings

 In this article, we will give you a guide to boosting your SEO rankings. 

Website optimization is the process of improving your online rankings with search engines to drive more customers to your website. Use the tips given below:

 Useful and consistent content: Useful content means providing information that keeps the user on your website longer. This helps to attract more customers to your website. You can publish blogs and videos to keep your content consistent. Add images to your content to optimize the content. 

 Page Speed: The slower your website loads, the more your bounce rate will skyrocket. Faster load speed leads to a better website experience means more website visits. More website visits signal to search engines that your website has got it going on. With Google, the user’s experience comes first, and slow load times mean dissatisfied users. Choose a company with a guarantee of at least 99.9% uptime. Ensure that your files are formatted and compressed correctly so they don’t swallow up your speed while loading. Avoid using too many plugins. The lack of cache can make your website slow. Ensure HTML text codes are clear.

 Internal Links: Internal links help with the navigation of a website and build equity in your site. Search engines find your posts and pages best when they’re linked to and from somewhere. It helps Google get a better idea of the structure of your website and allows for pages that are deeper within your site to rank better for related terms.

 Incorporate your location: Location-based SEO, also known as geo-targeting, is the practice of optimizing your content to appear in your target location. Google prioritizes local rankings based on 3 factors: relevance, distance, and prominence. 

 Optimize your website for local searches by completing your local directory profiles, using target location-specific keywords, and separating location pages.

 On-page user experience: Create a separate page for each location and publish content that is unique to each location. 

 So, you can bring more traffic to your website by optimizing the website using the above-given factors. Click here for more information https://fusiononemarketing.com/website-optimization/

Visit our social media pages:

https://www.facebook.com/fusiononemarketing

https://twitter.com/fusiononeteam

https://www.youtube.com/user/fusiononemarketing

https://www.linkedin.com/company/fusion-one-marketing/

https://www.google.com/maps?cid=3371799425934360683

[thedatasciencehyderabad]

data science in hyderabad

The Info Science

This Data Science course will make you skillful in building Recommender Engines, Predictive Models, performing Market Basket Analysis by making you're employed on live capstone initiatives. This Data Science Course in Hyderabad program allows you to master the real-world expertise in Data Science from the arms of expert Data Scientists. The corporations are at all times in hunt of licensed professionals who can carry out the evaluation at an excellent pace and showcase the initiatives they've worked on.

They provided me all the details concerning the course and payment construction. It's a every day basis class and I already attended 3 weeks courses, the college over-there's good and I am happy with the category surroundings. I paid a nominal cost of Rs. 9000 and I am satisfied with their teaching expertise. Satisfied with their training and response, For Artificial Intelligence in Data Science, Enrolled on this Omnitech Infosolutions, Ghatkesar, Hyderabad.

I would like to convey on behalf of the panellists that we discovered a lot of the students very good at communication and their technical abilities were also at par with what is expected based mostly on the curriculum offered to them. All the students seemed to be focussed about their profession objectives, had a pleasing angle to work together with and have been open to learn new things. Clustering - K-Means & Hierarchical Clustering is an unsupervised learning approach involving the grouping of knowledge. In this module, you will learn every little thing you have to know in regards to the technique and its varieties, like K-means clustering and hierarchical clustering. KNN KNN or k-Nearest Neighbours algorithm is likely one of the most easy machine studying algorithms for fixing regression and classification problems. This module will let us get snug with the Python programming language used for Data Science. We start with a high-stage idea of Object-Oriented Programming and later study the important vocabulary(/key phrases), grammar(/syntax) and sentence formation(/usable code) of this language.

The mentorship sessions with business specialists helped me turn into trade ready. The journey from an economics background to being an information analyst has been really wonderful. The program is very properly structured and an ideal mixture of concept and hands-on apply. Thanks to the DSE program at Great Lakes, I received 2 job provides, one from DXC Technology and another from Razorthink. This program is a perfect mix of both concept and palms-on practice. Taking this course to upskill myself was top-of-the-line selections I’ve made. Get an skilled profession mentor personalised to your academic background and experience.

They take care of huge quantities of data and make predictions utilizing statistical ideas. They should formulate and write queries and derive information from uncooked data. R is a programming language and free software environment for statistical computing and graphics supported by the R Foundation for Statistical Computing. Data science in R is done by utilizing packages like dplyr, ggplot2. It throws gentle on the prospects of a flourishing career in Analytics and what it takes to climb your approach to the top. This course not solely designed to enable you for brand new profession alternatives but additionally allows you to apply the new age abilities in your current work and turn out to be useful to in your current role.

It is a self-paced program of 48 credits that can be taken over 3-5 years. Similar to EMDS programme, college students will do 24 credit of coursework in the first two years. In the third 12 months, they'd two Capstone tasks of 12 credit each. If a pupil is not excited about doing the 24 credits of Capstone projects, then he/she will graduate with the Executive MTech in Data Science diploma.

In the continued pandemic, I researched for few online programs that I can do and I got here throughout InventaTeq. I joined the course and accomplished the Data Science certification.

If u r NON-IT background I advocate innomatics training institute to everyone. The instructing employees is good, they give us day wise assignments which helped me to hands on algorithms of machine learning and likewise provide us the backup classes. The entry they supply is very helpful to pay attention the courses repeatedly. ACTE is one of the best training institute for Data science and Data Analytics in BTM Layout. The trainers are properly skilled and the methodology of instructing is high notch. They provide practicals together with theoretical classes for complete understanding of the ideas. They even provide placement assistance after course completion as nicely.

Our Data Science course timings are nicely applicable for both working professionals & in addition to job seekers. With intense information & experience in Data Science area, you possibly can apply for the excessive paying job roles like Business Analyst, Data Engineer/Data Architect, Machine Learning Engineer, Big Data Engineer. Of course yes, we shall be scheduling backup classes for the scholars who have missed any of the important ideas on this Data Science coaching. InventaTeq has a various pool of scholars ranging from IT, Ecommerce, Business, Logistics and so on. Many of the trained professionals have turned themselves as an entrepreneur to help others in providing jobs and serve the nation.

This module will educate you about Probability Distributions and numerous varieties like Binomial, Poisson, and Normal Distribution in Python. Iterators Iterators are objects containing values, the place you can traverse through them. This module will educate you the way to implement Iterator strategies like __iter__() and __next__(). Loops A loop is used for executing a number of statements several times talked about. It will repeat the sequence of knowledge until it meets a specified situation. This module will teach you how to implement for and whereas loops.

click here for data science in hyd

Data Engineer –with a median wage of $151,498, to perform actual-time processing on information that is visualized and saved. Industry Architect –with a mean salary of $126,353, to investigate the enterprise system and optimize accordingly to support the development of updated technologies and system requirements. Applications Architect, whose average salary is $134,520, tracks functions conduct and applied in the enterprise to analyze the way they work together with the consumer. Data Architect, whose common salary is $137,630, builds knowledge solutions that may be utilized on a number of platforms. Business Intelligence Developer, with a median wage of $89,333, to design and develop business strategies for fast choice making and progress.

Their training is very good, they quoted Rs 20K for this coaching. They have been began a class for me a couple of days back, It's regular periods for 1 to 1.5 hours per day, very experienced trainers got via this academy for this AI course. Data Science program is really helpful for individuals who are looking for a career shift or wanting to grow in their analytics career. No, PAT does not promise a job, however it helps the aspirants to build the required potential needed in landing a career.

I saved looking out many institutes, after which I found Digital Nest, I approached and attended data science training demo, after which I realized that this is one of the finest coaching institutes that I should choose. That's the place I began studying Data Science training at the Hitech city branch of Hyderabad. I must say that I'm very a lot happy with the trainers and curriculum.

Extract helpful knowledge from web sites by performing web scraping using Python. Gain an in-depth understanding of Data Science processes, information wrangling, knowledge exploration, information visualization, speculation building, and testing. Demand for Data Scientist will zoom to twenty-eight% by 2020 across all business verticals.

I mainly wished to take up this course to step up in my group, so I cross checked the curriculum that digital nest offered with the my superiors. Even they were impressed and that really gave me the boldness to take up digital nest's data science certification course. Besides the above-talked about functions, the sphere of knowledge science provides outstanding advantages contributing so much to the sector it has been applied. Hence, many firms are contemplating data science to be an element to make their methods intelligent and derive essentially the most accurate outcomes. knowledge science has obtained huge advantages from time consumption to value reduction. Data science can be anticipated to exchange many current job roles.

[crookedtreepoetry]

Data Science Programs

You can enroll and complete the course to earn a shareable certificate, or you'll be able to audit it to view the course materials at no cost. When you subscribe to a course that's a part of a Certificate, you’re automatically subscribed to the total Certificate. Here, we’ll familiarise you with the basics of Data Science, MS Excel and other relevant ideas. We additionally offer a complimentary Python Programming Bootcamp and a profession important gentle expertise program with this course. Our course’s ultimate section introduces you to superior knowledge science concepts and covers topics similar to business intelligence, natural language processing, information engineering, and so forth.

One of essentially the most uncomfortable issues about studying data science on-line is that you just by no means really know if you’ve learned enough. Unlike in a formal college environment, when learning online you don’t have many good barometers for success, like passing or failing exams or entire courses. Projects assist remediate this by first displaying you what you don’t know, after which serving as a record of knowledge when it’s accomplished. The teacher makes this course actually fun and fascinating by supplying you with mock consulting projects to work on, then going by way of a complete walkthrough of the answer.

Students are properly prepared for getting recruited by the top corporations and work efficiently with different job profiles in Data Science. Subjects such as Analytics and Data Science have emerged as the most important needs of every company everywhere in the world. For every exercise that a business performs must be later analysed by information science professionals to extract valuable info. Jain University Entrance Test– This widespread entrance examination is carried out by Jain University, Bangalore to confess students who're excited about pursuing a bachelor’s diploma in information science. Data Science could be outlined as a mixture of implementing varied scientific activities corresponding to mathematics, calculus, graphs, charts, algorithms, computer packages, and much more.

Data science is a "concept to unify statistics, information analysis, informatics, and their associated methods" in order to "understand and analyze actual phenomena" with knowledge. It uses strategies and theories drawn from many fields within the context of mathematics, statistics, computer science, data science, and area knowledge. However, data science is totally different from pc science and data science. Our Master of Science in Data Science is the one on-line MSc program in information science.

Take this course when you’re uncomfortable with the linear algebra and calculus required for machine studying, and also you’ll save some time over other, extra generic math programs. Created by Andrew Ng, maker of the famous Stanford Machine Learning course, this is likely one of the highest rated knowledge science courses on the internet. Python is used in this course, and there’s many lectures going by way of the intricacies of the various knowledge science libraries to work through real-world, attention-grabbing issues. This is likely one of the solely knowledge science programs around that really touches on every a part of the info science process.

You will get job alternatives as soon as they're posted, recommendations to apply matched on to your skills and pursuits, and tips and tips that will help you stand other than the group. Each course sometimes incorporates 3-6 modules with a mean effort of 2 to four hours per module. If learning part-time (e.g. 1 module per week), it might take 6 to 12 months to finish the complete certificate. If learning full-time (e.g. 1 module per day) the certificates could be accomplished in 2 to 3 months. This Specialization is for software engineers, college students, and researchers from any subject, who are excited about machine learning and need to understand how GANs work.

Python is among the world’s most popular programming languages, and there has never been greater demand for professionals with the flexibility to use Python fundamentals to drive enterprise options throughout industries. This section of the course lasts for 12 weeks and consists of two assignments to check your data. We’ll introduce you to Python, Python programming, and the way you use Python in knowledge science.

With the expansion of the economic system of India and the growth of the enterprise sector, knowledge science has a fantastic scope in India. The college students must have accomplished their post-commencement in science with related specialisations corresponding to B.Sc. Statistics, B.Sc. Mathematics, B.Sc. Computer Science, or any other equal diploma from a recognised board or college.

In addition to earning a Professional Certificate from Coursera, you may additionally obtain a digital badge from IBM recognizing your proficiency in Data Science. They need a strong understanding of statistical analysis and arithmetic, that are important pillars of Data Science. A good understanding of these ideas will allow you to perceive the fundamental premises of data science. Machine studying is a priceless software to find patterns in large data units. To manage large data sets, data scientists have to be familiar with databases. However, non relational databases are rising in recognition, so a greater understanding of database structures is useful.

Only professionals having specialised information in the field of Data Science can perform these actions. This program is offered for people who're thinking about taking up the position of knowledge scientists however have no elementary thought about the area. Instead of focusing totally on a particular topic, the courses will stress on providing an summary of knowledge science, the types of issues tackled beneath it, and a few foundational strategies. By the ending, you should have a better sense of the way to proceed with extra advanced certification. There’s a base ability set and level of information that each one information scientists should possess, regardless of what trade they’re in.

It’s okay to complete just one course — you can pause your learning or end your subscription at any time. Visit your learner dashboard to track your course enrollments and your progress. This course offers an introduction to information science with no coding involved. In this non-technical course, you’ll be launched to every little thing you had been ever too afraid to ask about this quick-rising and thrilling area, while not having to put in writing a single line of code. In this course, you’ll learn to use the information, how to put together the information so as to perform operations on it. You’ll perceive the whole course of to extract important info from the pile of knowledge.

Besides this, corporations are able to pay good salaries to data analysts. This course is intended for Junior-stage Data Science students, and can allow them to establish, define, and explore numerous potential issues and initiatives, for comply with-up in the capstone course sequence. is a participant in the Amazon Services LLC Associates Program, an affiliate promoting program designed to supply a way for sites to earn promoting charges by advertising and linking to amazon.com.

From Predictive Analytics using Python to Machine Learning and from information visualization to Big Data everything is roofed in the comprehensive course. I hope these FREE Data Science Courses on 360digitmg will certainly help you to reinforce your knowledge of science and machine learning skills. If you've any doubt or questions, feel free to ask me within the remark section. After attempting a bunch of free on-line sources to be taught python, a colleague benefited from 360digitmg.

Data science seeks to seek out patterns in knowledge and use those patterns to predict future data. It draws on machine studying to course of giant quantities of information, discover patterns, and predict developments. It attracts from many scientific fields, and as a science, it progresses by creating new algorithms to analyze information and validate current methods. It is among the many best knowledge science courses for working professionals because it’s completely on-line and doesn’t require you to quit your job for continuing your studies. You will receive 25 expert coaching periods for doubt decision and progress feedback. This course sequence is likely one of the most enrolled in and highly rated course collections on this record.

I found courses, books, and papers that taught the things I wished to know, and then I utilized them to my project as I was learning. I discovered so much in a such short time period that it looks like an improbable feat if laid out as a curriculum. When becoming a member of any of those courses you need to make the same commitment to studying as you would in direction of a college course.

Yes, 360digitmg is sweet for brief courses that present sufficient info in a short period of time. 360digitmg is an Inexpensive, simple-to-use platform and has extra particular courses on information science. This Machine Learning Training will also assist you to understand the ideas of Statistics, Time Series, and totally different lessons of machine learning algorithms like supervised, unsupervised and reinforcement algorithms. Throughout the Machine Learning Training, you’ll be fixing actual-life case studies on Media, Healthcare, Social Media, Aviation, HR. Self-guided (but structured!) studying paths fully empower your solo research sessions day after day, but there's no purpose to do itall alone.

This is an entry-degree course so don’t count on to get detailed information about the topic. This is the basic step that you need to take so as to lead a successful data analyst profession. You will also know what areas are essential in Data Science and therefore could make informed selections on what areas to concentrate on learning. You can be a part of a couple of these programs to be taught Data Science from scratch and add an in-demand skill to your resume.

Since the primary course in this collection doesn’t spend any time instructing primary Python ideas, you must already be comfy with programming. Spending a while going via a platform like Treehouse would in all probability get you in control for the primary course. An extremely extremely rated course — 4.9/5 on SwichUp and four.8/5 on CourseReport — which is taught live by a knowledge scientist from a prime company.

Python being some of the broadly used languages is the brand new mantra for success. It is the primary tool for analytical professionals and is likely one of the high programming languages within the year 2019. Our purpose is to make the students get acquainted with python and become proficient in the most well-liked programming language.

It is likely one of the greatest options for science graduate students to go for pursuing a Master’s in Data Science after finishing their commencement. Data Science is a subject of study that offers the gathering, analysis, and processing of numerous information or info to extract options. It offers the understanding of many structured and unstructured data with specialised information to get the required insights.

By completing the assignments and lectures, you'll earn the certification of completion, which could be showcased in job purposes. If you’re extra involved within the machine studying aspect of information science, take a look at the Top 5 Machine Learning Courses for 2020 as a complement to this text. When I first began studying information science and machine learning, I began by making an attempt to predict stocks.

If you plan on taking this course it would be a good idea to pair it with a separate statistics and likelihood course as well. With an excellent mixture of theory and software, this course from Harvard is one of the finest for getting began as a newbie. It’s not on an interactive platform, like Coursera or edX, and doesn’t supply any type of certification, however it positively prices your time and it’s totally free. The one downside of this MicroMaster’s, and lots of courses on 360digitmg , is that they aren’t offered as incessantly as different platforms.

Explore more on - data science course in hyderabad with placements

360DigiTMG - Data Analytics, Data Science Course Training Hyderabad

 Address:-2-56/2/19, 3rd floor, Vijaya towers, near Meridian school, Ayyappa Society Rd, Madhapur, Hyderabad, Telangana 500081

Contact us ( 099899 94319 )

Hours: Sunday - Saturday 7 AM - 11 PM

[politicalargumentativeessay866]

essay writers

About me

Essay Help Online You Can Use Anonymously

Essay Help Online You Can Use Anonymously Plan ahead to keep away from writing your essay in a hurry. If something goes wrong and you have a very restricted period of time to complete your project, you'll have to sacrifice something. Miss some phases of writing it - like version and making a top level view. First, you need to have schooling and expertise in your area. We insist upon both of these after we rent writers. In larger classes, your professor will doubtless be assisted by a TA, or instructing assistant. This is likely a graduate pupil whose job is to grade assignments and supply students with help. You specify what sort of essay you need help with, flick through the listing of writers prepared to start out working on your paper and choose the one you like. You can rely on their expertise and background and ask to see a few of the essay samples. A team of vetted diploma-holders and skilled essay writers is here for you. As a college student, lab stories, articles, critiques, case research, and a variety of different writing assignments will come your method. Most students full most of these assignments with little to no exterior assist. However, you never need to do it alone, as a result of we are there each and every time. Anytime you have an active project with us, you possibly can interact together with your author instantly all through this course of. Just check in to your customer account, then send and obtain messages with your author. In this fashion, you realize the precise standing of your project from the time you place your order until the ultimate project is completed. There are two major traits of a profitable tutorial author. They may even lead a category or two all through the semester. Your TA is there to help students who ask for it, and it’s by no means a nasty concept to go to them for advice. If you possibly can’t discover a class by way of your school, and you can’t afford to pay for a course, you can research this topic on your own. Additionally, should you flip to the internet for writing suggestions, be ready to spend a big chunk of time sifting via data to find a respected supply. We will take your essay and make it move like Hemingway. Our writers are highly expert at catching even the slightest error. Let's also think about some common tips on tips on how to find yourself getting the grade you want on your essay. The first obvious one is to do every little thing you can to have as a lot time as you should write it properly. EssayDog® is a software that helps stroll students like you through the process of writing faculty essays. EssayDog® can help you establish the elements of a great story which you can then use to create a fantastic essay. Students who use EssayDog® watch quick videos that guide them via the process of brainstorming and constructing an essay, then complete a brief writing train after every video. These workout routines will allow you to pinpoint your story in four sentences, find the overarching message and theme of your story, and hyperlink your story to the essay prompt. EssayDog® is a great resource for college students who wrestle with essay construction. It will make your essay not as polished and spectacular however at least it is possible for you to to show it in by the deadline. EduPeet.com is the place where you can shortly find a talented essay writer to assist you. The algorithm for getting the perfect ultimate result's straightforward. The specialists working at our company hold degrees in relevant fields of knowledge. They know the way to use varied citation kinds and have handed numerous tests to show their skills. They do their finest to earn the approval of students. Those who use our service know they can rely on the specialists of their alternative. Our writers have been completely checked to seek out out whether or not they meet our requirements. If you resolve you want an essay ASAP and the timing is already past working hours, you don’t have to fret. Just contact our buyer support department through stay chat, and they'll instantly let you understand if an acceptable author is on the market to supply help to write down an essay. We don’t distribute the tasks on random order. When a student asks help with essay writing, we make sure to pair them with a properly trained writer within the applicable area of research.

[persuasiveessaywritingprompts241]

write my research papers

About me

Why Students Should Still Write Research Papers

Why Students Should Still Write Research Papers If you are feeling your presentation abilities still need improvement you may ask your self can I pay somebody to write down my paper? The content is your individual and you have gained the knowledge on the subject that you'll be taking ahead. Get someone to write my paper as a result of it's not plagiarism. Ask an company are you able to write a paper for me and in case you have chosen the proper one, the answer will be within the affirmative. Students of English could also be much less likely to want assistance, but it's there if wanted. Asking an company to help me write my paper remains to be perfectly acceptable. The point about training is to gain data in your chosen topic so as to observe your chosen career. That knowledge is an important level to keep in mind at all times. Your capability to present it to examiners is why you could ask your self who might help me with my paper? While writing the research paper, keep in mind that your work might be learn not just by scientists concerned in the subject but by ordinary individuals. Use simple words and don't deny to make use of a simple sentence. Otherwise, the reader will be misplaced in too long and complex text. Devote at least one sentence per paragraph for an argument. Hire somebody to write down a paper in order that your information translates into a well-written presentation. I am a component-time scholar presently pursuing a level course and working full-time. Hence, it leaves me with little time to do a properly-written research paper. Authentic papers written from scratch per your directions, free plagiarism report. If you realize the main steps from the thought to research writing to be made, this could possibly be of a lot research paper help for you. Writing my papers this 12 months has been a wrestle, what with all these midterms and work. Now, barring some type of a zombie apocalypse, a minimum of I know my papers are taken care of haha. It is not cheating as a result of you're the originator of the content material and the arguments and conclusions that the paper puts forward. The presentation of that research may need apply, even assist. A good company offers loads of info on its web site for you to read earlier than you determine to make contact. Once you're feeling comfortable with what you've learn, you have to fill in some fundamental info to start out the method of getting the writer you want. Similarly, these having fun with the historical past and desirous to convey that like to others want to have the ability to get their ardour throughout to others. Do not overlook to look up for an argument in works you’ve cited to write a analysis paper. On this stage, it's essential to select the obtainable results of comparable researches and take out the redundant data. It enables you to ask for exact info you need. Be artistic, try to find glorious books that completely fit your subject. There is an express algorithm for research paper writing. Don't you just adore it once they assign you a 10-web page project throughout midterms? Then let our academics construct you a foolproof paper after hours of research. Yes, there are three companies any student will find useful. Along with the chance to decide on a writer’s class and to check his or her fashion with samples, there are some other options you should use. If even after all of the modifications and revisions you’re still unsatisfied together with your paper, contact our high quality assurance division to apply for a refund.

[responseessayexamples171]

write essay for me

About me

Essay Writer & Essay Writing Service

Essay Writer & Essay Writing Service After you compose the essay by yourself it is sti have to go to the library to analysis in your favorite theme to debate, and so on… The article must be all finished with summary. At times you might need recommendations therefore it's potential to be in a spot to compose and finish your personal essay paper about the best way to compose a article. I obtained 3-web page argumentative essay overnight from the TOP writer. It was dearer than basic writer, however I'm totally glad with the end result. With us, you can stay fully nameless while purchasing killer essays. If your educational challenge is pressing enough, we will end the piece in only 3 hours. Our products are authentic and free from plagiarism. All of our analysis sources are dependable and compatible along with your needs. A plagiarism software program is used to ensure that our prospects receive a hundred% distinctive college essays. In fact, it requires the information of the associated subject and sure stage of creativity. Writers working for MeowEssay are committed to their shoppers’ orders and to delivering exceptional tutorial papers in accordance with all necessities. The ideal biography article will produce curiosity in your own reader this fashion. So are you too busy to catch a break from a busy day? Are you hellbent on making one of the best out of your faculty life but simply lack time to do so? Then we're happy that you simply’ve discovered us, as we’re the platform looking for to help. Managers make clear all the main points of your order after which can care for explaining every thing to the author. I did not actually have to do anything but pay after which obtain accomplished essay. Essay is the sort of academic writing aiming to construct your paper in accordance with some specific subjects. Essay writing is a tricky task that requires much effort and dedication. Or let’s fancy a dream party with all your folks around. Everything is okay, everyone’s pleased but then… something is popping up and also you say, ‘I must do my essay’. You usually are not certain of how to current enough evidence to make your argument worthy in essay writing. We can make gentle work of essay writing for you – just ship us your directions. You can have your paper revised up to 3 times on the sweet worth of zero. We prolong this supply as a result of we understand how essential it is for you to have an ideal essay. We persistently search for new avenues of reducing our costs, which is why each you and your wallet will be pleased concerning the selection of this writing service. All evening is ruined, individuals get upset, you urgently get down to work. If you don't like this state of affairs, it is preferable to save lots of precious time and good mood. We need you to have control over your on-line identity. To this end, we employ multilayered encryption algorithms and proprietary safety protocols. You can finally give us all your troubles and luxuriate in a great evening’s relaxation or some quality time along with your mates. No, this information is confidential, and nobody can see it, not even your writer. In case you aren't comfortable having some personal knowledge on your account, we will delete it at your request. In any case, no one can use our system to study some private details about you or find out you used this service. We usually are not like all different websites that promise you the most effective, but provide you with providers of students from third world countries with poor data. Our staff of training consultants help you in your particular person communication as long as you want it. We work on comments and recommendations to improve writing and performance of our customized essays. It just isn't needed that each essay flows properly, as everybody just isn't gifted with distinctive writing expertise. People who aren't that creative need to not fear as a result of British Essay Writers are here to help them improve their work. As we understand the wish of every pupil is to seek out somebody who can work immediately on their request of “write my essay online UK”.

[thebottlehoarder]

Preston SEO Service – Your seo Agency

Take advantage of the Best SEO Preston Services

Take advantage of the Best SEO Preston Services

As an online business owner, you already know how crucial it is to have a strong web presence to compete in the industry. However, the way search engines rank websites are constantly changing. What works before may not be applicable today or in the coming days, weeks or months. You need to remain up-to-date to keep up with these changes and ensure a high ranking on search engine results pages (SERPs). You need to use the right approach, so that search engines like Google, will recognize your website as something relevant and valuable to customers. This is where SEO Preston comes in.

If you want to stay ahead of the competition, you need to integrate the best SEO technique to be found in organic search results. Whether you have a new business site or looking to increase your online visibility, a reliable SEO agency can be of paramount help. SEO Preston is here to help catch the attention of your potential customers looking for your business, maintain a strong web presence and increase your online traffic.

Digitery – Your Ultimate Partner in Excellent SEO Preston

There’s no other form of digital marketing today that can deliver high ROI quite like search engine optimization. However, not all SEO companies in Preston are the same. Some claim to provide high-end SEO strategies but fail to deliver the best results you need. Others say that they can help you succeed online but don’t really have what it takes to take you on top of the industry.

Fortunately, Digitery is now here to deliver outstanding results based on its strategic methodologies and state-of-the-art techniques. With a team of highly-trained and experienced SEO professionals, Digitery is equipped to help you achieve more traffic and increase your popularity online. We are employed with the most reputable digital marketing experts with in-depth knowledge of Google algorithm. They know how to help major search engines find your website and trust you.

We are confident with our service. This is because each of our search engine marketing approach is backed by many years of online experience that allow us to fully understand the industry and help you succeed, regardless of your business. We have satisfied customers who can affirm that we are the ultimate company to trust when looking for a reliable SEO Preston.

Why Choose Us?

You’ve probably heard so many times that you need search engine optimization. If you are looking to have more online visitors and increased conversions, then you should definitely choose the best SEO company in Preston like Digitery. Consistent online exposure is crucial if you really want to grow your business. Allow us to take the burden off your shoulder. Digitery will take your business to higher levels of success through the following:

Acquire New Clients and Keep the Existing Ones. Your goal as a business owner is not only to get new customers who buy your products and services. You also need to focus on ensuring that your new clients will come and continue to buy. Digitery is pleased to offer unique website optimization services like PPC, social media management, content creation, link building and more. We ensure to integrate the best strategy that works specifically for your business to obtain the fastest sales.

Gain more Local Customers. Why go too far when you can easily target and reach your local customers in Preston? Our unique optimization services are designed to help your website become visible to people who are searching for your business in the local area. We ensure that your website obtain a first-page placement, establish strong reputation and increase exposure through cost-effective SEO techniques.

Effective Branding. One of the best things about SEO Preston is that it does not just help your website become visible online, but also help boost your brand. With the right strategy, it can deliver consistent and endless traffic to your site, which is vital in improving your brand. You will be recognized in the industry and customers will continue trusting you.

SEO Preston is Cost-Efficient. Are you looking for an effective way to increase your business exposure without breaking the bank? Do you need help improving your conversion but can’t afford those extravagant marketing services in the market? Don’t worry! You will never go wrong with SEO services in Preston. We have the most skilled internet marketers and SEO specialists who are dedicated to reaching your goals without spending too much. We deliver the best results you need at cost-effective prices.

Don’t look any further when searching for a reputable SEO specialist in Preston. Digitery is here to help you in every step of the way.

Outstanding SEO Services in Preston

Don’t exactly know which SEO service to use for your business? To get started, here are among the most outstanding SEO services you can take full advantage of in Preston.

Website Optimization and Content Development

You need quality content to be recognized by search engines. It must be relevant, useful, unique and fresh. Most of all, it should transform your traffic into customers that pay. Allow our content marketing specialists to write high-quality content that matches your business.

Keyword Search

We will determine the best keywords and keyword phrases that will get you on top of SERPs. Our SEO experts will conduct a thorough research to identify which keywords will best help optimize your website.

Local Search Optimization

Looking to reach those people in the area who are looking for your business? Your local customers are ready to buy. All you have to do is to appear in front of them when they look for your services or products and grab their attention. We have the most experienced SEO professionals to help you with that.

Web Marketing Analytics

SEO experts will establish analytics reporting on your business site and deliver updated consultations and recommendations to help you use the information obtained on your advantage. All related data and metrics in relation to your visitors will also be determined to know which keywords drive high traffic.

Search Engine Ranking Monitoring

Having knowledge where your website ranks on search engines will help you determine the areas that need improvement to reach success. Monitoring your search engine ranking will also help you identify your strengths and which areas of marketing you need to work on.

SEO Analysis and Audit

We will review and assess your website for all online and offsite factors that affect your online traffic. The results will be presented in an understandable manner and with clear recommendations and possible solutions for improvement.

Digitery is your ultimate partner when looking for the best SEO Preston. With many years of industry experience and dedication to customer care, we ensure to deliver the best digital marketing services to grow your business.

Source: https://digitery.co.uk/website-design-app-development/seo-preston/

[doorrepcal33169]

You created a machine learning application. Now make sure it’s secure.

The software industry has demonstrated, all too clearly, what happens when you don’t pay attention to security.

In a recent post, we described what it would take to build a sustainable machine learning practice. By “sustainable,” we mean projects that aren’t just proofs of concepts or experiments. A sustainable practice means projects that are integral to an organization’s mission: projects by which an organization lives or dies. These projects are built and supported by a stable team of engineers, and supported by a management team that understands what machine learning is, why it’s important, and what it’s capable of accomplishing. Finally, sustainable machine learning means that as many aspects of product development as possible are automated: not just building models, but cleaning data, building and managing data pipelines, testing, and much more. Machine learning will penetrate our organizations so deeply that it won’t be possible for humans to manage them unassisted.

Organizations throughout the world are waking up to the fact that security is essential to their software projects. Nobody wants to be the next Sony, the next Anthem, or the next Equifax. But while we know how to make traditional software more secure (even though we frequently don’t), machine learning presents a new set of problems. Any sustainable machine learning practice must address machine learning’s unique security issues. We didn’t do that for traditional software, and we’re paying the price now. Nobody wants to pay the price again. If we learn one thing from traditional software’s approach to security, it’s that we need to be ahead of the curve, not behind it. As Joanna Bryson writes, “Cyber security and AI are inseparable.”

The presence of machine learning in any organization won’t be a single application, a single model; it will be many applications, using many models—perhaps thousands of models, or tens of thousands, automatically generated and updated. Machine learning on low-power edge devices, ranging from phones to tiny sensors embedded in assembly lines, tools, appliances, and even furniture and building structures, increases the number of models that need to be monitored. And the advent of 5G mobile services, which significantly increases the network bandwidth to mobile devices, will make it much more attractive to put machine learning at the edge of the network. We anticipate billions of machines, each of which may be running dozens of models. At this scale, we can't assume that we can deal with security issues manually. We need tools to assist the humans responsible for security. We need to automate as much of the process as possible, but not too much, giving humans the final say.

In “Lessons learned turning machine learning models into real products and services,” David Talby writes that “the biggest mistake people make with regard to machine learning is thinking that the models are just like any other type of software.” Model development isn’t software development. Models are unique—the same model can’t be deployed twice; the accuracy of any model degrades as soon as it is put into production; and the gap between training data and live data, representing real users and their actions, is huge. In many respects, the task of modeling doesn’t get started until the model hits production, and starts to encounter real-world data.

Unfortunately, one characteristic that software development has in common with machine learning is a lack of attention to security. Security tends to be a low priority. It gets some lip service, but falls out of the picture when deadlines get tight. In software, that’s been institutionalized in the “move fast and break things” mindset. If you’re building fast, you’re not going to take the time to write sanitary code, let alone think about attack vectors. You might not “break things,” but you’re willing to build broken things; the benefits of delivering insecure products on time outweigh the downsides, as Daniel Miessler has written. You might be lucky; the vulnerabilities you create may never be discovered. But if security experts aren’t part of the development team from the beginning, if security is something to be added on at the last minute, you’re relying on luck, and that’s not a good position to be in. Machine learning is no different, except that the pressure of delivering a product on time is even greater, the issues aren’t as well understood, the attack surface is larger, the targets are more valuable, and companies building machine learning products haven’t yet engaged with the problems.

What kinds of attacks will machine learning systems see, and what will they have to defend against? All of the attacks we have been struggling with for years, but there are a number of vulnerabilities that are specific to machine learning. Here’s a brief taxonomy of attacks against machine learning:

Poisoning, or injecting bad (“adversarial”) data into the training data. We’ve seen this many times in the wild. Microsoft’s Tay was an experimental chatbot that was quickly taught to spout racist and anti-semitic messages by the people who were chatting with it. By inserting racist content into the data stream, they effectively gained control over Tay’s behavior. The appearance of “fake news” in channels like YouTube, Facebook, Twitter, and even Google searches, was similar: once fake news was posted, users were attracted to it like flies, and the algorithms that made recommendations “learned” to recommend that content. danah boyd has argued that these incidents need to be treated as security issues, intentional and malicious corruption of the data feeding the application, not as isolated pranks or algorithmic errors.

Any machine learning system that constantly trains itself is vulnerable to poisoning. Such applications could range from customer service chat bots (can you imagine a call center bot behaving like Tay?) to recommendation engines (real estate redlining might be a consequence) or even to medical diagnosis (modifying recommended drug dosages). To defend against poisoning, you need strong control over the training data. Such control is difficult (if not impossible) to achieve. “Black hat SEO” to improve search engine rankings is nothing if not an early (and still very present) example of poisoning. Google can’t control the incoming data, which is everything that is on the web. Their only recourse is to tweak their search algorithms constantly and penalize abusers for their behavior. In the same vein, bots and troll armies have manipulated social media feeds to spread views ranging from opposition to vaccination to neo-naziism.

Evasion, or crafting input that causes a machine learning system to misclassify it. Again, we’ve seen this both in the wild and in the lab. CV Dazzle uses makeup and hair styles as “camouflage against face recognition technology.” Other research projects have shown that it’s possible to defeat image classification by changing a single pixel in an image: a ship becomes a car, a horse becomes a frog. Or, just as with humans, image classifiers can miss an unexpected object that’s out of context: an elephant in the room, for example. It’s a mistake to think that computer vision systems “understand” what they see in ways that are similar to humans. They’re not aware of context, they don’t have expectations about what’s normal; they’re simply doing high-stakes pattern matching. Researchers have reported similar vulnerabilities in natural language processing, where changing a word, or even a letter, in a way that wouldn’t confuse human researchers causes machine learning to misunderstand a phrase.

Although these examples are often amusing, it’s worth thinking about real-world consequences: could someone use these tricks to manipulate the behavior of autonomous vehicles? Here’s how that could work: I put a mark on a stop sign—perhaps by sticking a fragment of a green sticky note at the top. Does that make an autonomous vehicle think the stop sign is a flying tomato, and if so, would the car stop? The alteration doesn’t have to make the sign “look like” a tomato to a human observer; it just has to push the image closer to the boundary where the model says “tomato.” Machine learning has neither the context nor the common sense to understand that tomatoes don’t appear in mid-air. Could a delivery drone be subverted to become a weapon by causing it to misunderstand its surroundings? Almost certainly. Don’t dismiss these examples as academic. A stop sign with a few pixels changed in the lab may not be different from a stop sign that has been used for target practice during hunting season.

Impersonation attacks attempt to fool a model into misidentifying someone or something. The goal is frequently to gain unauthorized access to a system. For example, an attacker might want to trick a bank into misreading the amount written on a check. Fingerprints obtained from drinking glasses, or even high resolution photographs, can be used to fool fingerprint authentication. South Park trolled Alexa and Google Home users by using the words “Alexa” and “OK Google” repeatedly in an episode, triggering viewers’ devices; the devices weren’t able to distinguish between the show voices and real ones. The next generation of impersonation attacks will be “deep fake” videos that place words in the mouths of real people.

Inversion means using an API to gather information about a model, and using that information to attack it. Inversion can also mean using an API to obtain private information from a model, perhaps by retrieving data and de-anonymizing it. In “The Secret Sharer: Measuring Unintended Neural Network Memorization & Extracting Secrets,” the authors show that machine learning models tend to memorize all their training data, and that it’s possible to extract protected information from a model. Common approaches to protecting information don’t work; the model still incorporates secret information in ways that can be extracted. Differential privacy—the practice of carefully inserting extraneous data into a data set in ways that don’t change its statistical properties—has some promise, but with significant cost: the authors point out that training is much slower. Furthermore, the number of developers who understand and can implement differential privacy is small.

While this may sound like an academic concern, it’s not; writing a script to probe machine learning applications isn’t difficult. Furthermore, Michael Veale and others write that inversion attacks raise legal problems. Under the GDPR, if protected data is memorized by models, are those models subject to the same regulations as personal data? In that case, developers would have to remove personal data from models—not just the training data sets—on request; it would be very difficult to sell products that incorporated models, and even techniques like automated model generation could become problematic. Again, the authors point to differential privacy, but with the caution that few companies have the expertise to deploy models with differential privacy correctly.

Other vulnerabilities, other attacks

This brief taxonomy of vulnerabilities doesn’t come close to listing all the problems that machine learning will face in the field. Many of these vulnerabilities are easily exploited. You can probe Amazon to find out what products are recommended along with your products, possibly finding out who your real competitors are, and discovering who to attack. You might even be able to reverse-engineer how Amazon makes recommendations and use that knowledge to influence the recommendations they make.

More complex attacks have been seen in the field. One involves placing fake reviews on an Amazon seller’s site, so that when the seller removes the reviews, Amazon bans the seller for review manipulation. Is this an attack against machine learning? The attacker tricks the human victim into violating Amazon’s rules. Ultimately, though, it’s the machine learning system that’s tricked into taking an incorrect action (banning the victim) that it could have prevented.

��Google bowling” means creating large numbers of links to a competitor’s website in hopes that Google’s ranking algorithm will penalize the competitor for purchasing bulk links. It’s similar to the fake review attack, except that it doesn’t require a human intermediary; it’s a direct attack against the algorithm that analyzes inbound links.

Advertising was one of the earliest adopters of machine learning, and one of the earliest victims. Click fraud is out of control, and the machine learning community is reluctant to talk about (or is unaware of) the issue—even though, as online advertising becomes ever more dependent on machine learning, fraudsters will learn how to attack models directly in their attempts to appear legitimate. If click data is unreliable, then models built from that data are unreliable, along with any results or recommendations generated by those models. And click fraud is similar to many attacks against recommendation systems and trend analysis. Once a “fake news” item has been planted, it’s simple to make it trend with some automated clicks. At that point, the recommendation takes over, generating recommendations which in turn generate further clicks. Anything automated is prone to attack, and automation allows those attacks to take place at scale.

The advent of autonomous vehicles, ranging from cars to drones, presents yet another set of threats. If the machine learning systems on an autonomous vehicle are vulnerable to attack, a car or truck could conceivably be used as a murder weapon. So could a drone—either a weaponized military drone or a consumer drone. The military already knows that drones are vulnerable; in 2011, Iran captured a U.S. drone, possibly by spoofing GPS signals. We expect to see attacks on “smart” consumer health devices and professional medical devices, many of which we know are already vulnerable.

Taking action

Merely scolding and thinking about possible attacks won’t help. What can be done to defend machine learning models? First, we can start with traditional software. The biggest problem with insecure software isn’t that we don’t understand security; it’s that software vendors, and software users, never take the basic steps they would need to defend themselves. It’s easy to feel defenseless before hyper-intelligent hackers, but the reality is that sites like Equifax become victims because they didn’t take basic precautions, such as installing software updates. So, what do machine learning developers need to do?

Security audits are a good starting point. What are the assets that you need to protect? Where are they, and how vulnerable are they? Who has access to those resources, and who actually needs that access? How can you minimize access to critical data? For example, a shipping system needs customer addresses, but it doesn’t need credit card information; a payment system needs credit card information, but not complete purchase histories. Can this data be stored and managed in separate, isolated databases? Beyond that, are basic safeguards in place, such as two-factor authentication? It’s easy to fault Equifax for not updating their software, but almost any software system depends on hundreds, if not thousands, of external libraries. What strategy do you have in place to ensure they’re updated, and that updates don't break working systems?

Like conventional software, machine learning systems should use monitoring systems that generate alerts to notify staff when something abnormal or suspicious occurs. Some of these monitoring systems are already using machine learning for anomaly detection—which means the monitoring software itself can be attacked.

Penetration testing is a common practice in the online world: your security staff (or, better, consultants) attack your site to discover its vulnerabilities. Attack simulation is an extension of penetration testing that shows you “how attackers actually achieve goals against your organization.” What are they looking for? How do they get to it? Can you gain control over a system by poisoning its inputs?

Tools for testing computer vision systems by generating "adversarial images" are already appearing, such as cleverhans and IBM’s ART. We are starting to see papers describing adversarial attacks against speech recognition systems. Adversarial input is a special case of a more general problem. Most machine learning developers assume their training data is similar to the data their systems will face in the real world. That’s an idealized best case. It’s easy to build a face identification system if all your faces are well-lit, well-focused, and have light-skinned subjects. A working system needs to handle all kinds of images, including images that are blurry, badly focused, poorly lighted—and have dark-skinned subjects.

Safety verification is a new area for AI research, still in its infancy. Safety verification asks questions like whether models can deliver consistent results, or whether small changes in the input lead to large changes in the output. If machine learning is at all like conventional software, we expect an escalating struggle between attackers and defenders; better defenses will lead to more sophisticated attacks, which will lead to a new generation of defenses. It will never be possible to say that a model has been “verifiably safe.” But it is important to know that a model has been tested, and that it is reasonably well-behaved against all known attacks.

Model explainability has become an important area of research in machine learning. Understanding why a model makes specific decisions is important for several reasons, not the least of which is that it makes people more comfortable with using machine learning. That “comfort” can be deceptive, of course. But being able to ask models why they made particular decisions will conceivably make it easier to see when they’ve been compromised. During development, explainability will make it possible to test how easy it is for an adversary to manipulate a model, in applications from image classification to credit scoring. In addition to knowing what a model does, explainability will tell us why, and help us build models that are more robust, less subject to manipulation; understanding why a model makes decisions should help us understand its limitations and weaknesses. At the same time, it’s conceivable that explainability will make it easier to discover weaknesses and attack vectors. If you want to poison the data flowing into a model, it can only help to know how the model responds to data.

In “Deep Automation in Machine Learning,” we talked about the importance of data lineage and provenance, and tools for tracking them. Lineage and provenance are important whether or not you’re developing the model yourself. While there are many cloud platforms to automate model building and even deployment, ultimately your organization is responsible for the model’s behavior. The downside of that responsibility includes everything from degraded profits to legal liability. If you don’t know where your data is coming from and how it has been modified, you have no basis for knowing whether your data has been corrupted, either through accident or malice.

“Datasheets for Datasets” proposes a standard set of questions about a data set’s sources, how the data was collected, its biases, and other basic information. Given a specification that records a data set’s properties, it should be easy to test and detect sudden and unexpected changes. If an attacker corrupts your data, you should be able to detect that and correct it up front; if not up front, then later in an audit.

Datasheets are a good start, but they are only a beginning. Whatever tools we have for tracking data lineage and provenance need to be automated. There will be too many models and data sets to rely on manual tracking and audits.

Balancing openness against tipping off adversaries

In certain domains, users and regulators will increasingly prefer machine learning services and products that can provide simple explanations for how automated decisions and recommendations are being made. But we’ve already seen that too much information can lead to certain parties gaming models (as in SEO). How much to disclose depends on the specific application, domain, and jurisdiction.

This balancing act is starting to come up in machine learning and related areas that involve the work of researchers (who tend to work in the open) who are up against adversaries who prize unpublished vulnerabilities. The question of whether or not to “temporarily hold back” research results is a discussion that the digital media forensics community has been having. In a 2018

from FEED 10 TECHNOLOGY https://ift.tt/2Vrb0Ym

[csemntwinl3x0a1]

You created a machine learning application. Now make sure it’s secure.

You created a machine learning application. Now make sure it’s secure.

The software industry has demonstrated, all too clearly, what happens when you don’t pay attention to security.

In a recent post, we described what it would take to build a sustainable machine learning practice. By “sustainable,” we mean projects that aren’t just proofs of concepts or experiments. A sustainable practice means projects that are integral to an organization’s mission: projects by which an organization lives or dies. These projects are built and supported by a stable team of engineers, and supported by a management team that understands what machine learning is, why it’s important, and what it’s capable of accomplishing. Finally, sustainable machine learning means that as many aspects of product development as possible are automated: not just building models, but cleaning data, building and managing data pipelines, testing, and much more. Machine learning will penetrate our organizations so deeply that it won’t be possible for humans to manage them unassisted.

Organizations throughout the world are waking up to the fact that security is essential to their software projects. Nobody wants to be the next Sony, the next Anthem, or the next Equifax. But while we know how to make traditional software more secure (even though we frequently don’t), machine learning presents a new set of problems. Any sustainable machine learning practice must address machine learning’s unique security issues. We didn’t do that for traditional software, and we’re paying the price now. Nobody wants to pay the price again. If we learn one thing from traditional software’s approach to security, it’s that we need to be ahead of the curve, not behind it. As Joanna Bryson writes, “Cyber security and AI are inseparable.”

The presence of machine learning in any organization won’t be a single application, a single model; it will be many applications, using many models—perhaps thousands of models, or tens of thousands, automatically generated and updated. Machine learning on low-power edge devices, ranging from phones to tiny sensors embedded in assembly lines, tools, appliances, and even furniture and building structures, increases the number of models that need to be monitored. And the advent of 5G mobile services, which significantly increases the network bandwidth to mobile devices, will make it much more attractive to put machine learning at the edge of the network. We anticipate billions of machines, each of which may be running dozens of models. At this scale, we can't assume that we can deal with security issues manually. We need tools to assist the humans responsible for security. We need to automate as much of the process as possible, but not too much, giving humans the final say.

In “Lessons learned turning machine learning models into real products and services,” David Talby writes that “the biggest mistake people make with regard to machine learning is thinking that the models are just like any other type of software.” Model development isn’t software development. Models are unique—the same model can’t be deployed twice; the accuracy of any model degrades as soon as it is put into production; and the gap between training data and live data, representing real users and their actions, is huge. In many respects, the task of modeling doesn’t get started until the model hits production, and starts to encounter real-world data.

Unfortunately, one characteristic that software development has in common with machine learning is a lack of attention to security. Security tends to be a low priority. It gets some lip service, but falls out of the picture when deadlines get tight. In software, that’s been institutionalized in the “move fast and break things” mindset. If you’re building fast, you’re not going to take the time to write sanitary code, let alone think about attack vectors. You might not “break things,” but you’re willing to build broken things; the benefits of delivering insecure products on time outweigh the downsides, as Daniel Miessler has written. You might be lucky; the vulnerabilities you create may never be discovered. But if security experts aren’t part of the development team from the beginning, if security is something to be added on at the last minute, you’re relying on luck, and that’s not a good position to be in. Machine learning is no different, except that the pressure of delivering a product on time is even greater, the issues aren’t as well understood, the attack surface is larger, the targets are more valuable, and companies building machine learning products haven’t yet engaged with the problems.

What kinds of attacks will machine learning systems see, and what will they have to defend against? All of the attacks we have been struggling with for years, but there are a number of vulnerabilities that are specific to machine learning. Here’s a brief taxonomy of attacks against machine learning:

Poisoning, or injecting bad (“adversarial”) data into the training data. We’ve seen this many times in the wild. Microsoft’s Tay was an experimental chatbot that was quickly taught to spout racist and anti-semitic messages by the people who were chatting with it. By inserting racist content into the data stream, they effectively gained control over Tay’s behavior. The appearance of “fake news” in channels like YouTube, Facebook, Twitter, and even Google searches, was similar: once fake news was posted, users were attracted to it like flies, and the algorithms that made recommendations “learned” to recommend that content. danah boyd has argued that these incidents need to be treated as security issues, intentional and malicious corruption of the data feeding the application, not as isolated pranks or algorithmic errors.

Any machine learning system that constantly trains itself is vulnerable to poisoning. Such applications could range from customer service chat bots (can you imagine a call center bot behaving like Tay?) to recommendation engines (real estate redlining might be a consequence) or even to medical diagnosis (modifying recommended drug dosages). To defend against poisoning, you need strong control over the training data. Such control is difficult (if not impossible) to achieve. “Black hat SEO” to improve search engine rankings is nothing if not an early (and still very present) example of poisoning. Google can’t control the incoming data, which is everything that is on the web. Their only recourse is to tweak their search algorithms constantly and penalize abusers for their behavior. In the same vein, bots and troll armies have manipulated social media feeds to spread views ranging from opposition to vaccination to neo-naziism.

Evasion, or crafting input that causes a machine learning system to misclassify it. Again, we’ve seen this both in the wild and in the lab. CV Dazzle uses makeup and hair styles as “camouflage against face recognition technology.” Other research projects have shown that it’s possible to defeat image classification by changing a single pixel in an image: a ship becomes a car, a horse becomes a frog. Or, just as with humans, image classifiers can miss an unexpected object that’s out of context: an elephant in the room, for example. It’s a mistake to think that computer vision systems “understand” what they see in ways that are similar to humans. They’re not aware of context, they don’t have expectations about what’s normal; they’re simply doing high-stakes pattern matching. Researchers have reported similar vulnerabilities in natural language processing, where changing a word, or even a letter, in a way that wouldn’t confuse human researchers causes machine learning to misunderstand a phrase.

Although these examples are often amusing, it’s worth thinking about real-world consequences: could someone use these tricks to manipulate the behavior of autonomous vehicles? Here’s how that could work: I put a mark on a stop sign—perhaps by sticking a fragment of a green sticky note at the top. Does that make an autonomous vehicle think the stop sign is a flying tomato, and if so, would the car stop? The alteration doesn’t have to make the sign “look like” a tomato to a human observer; it just has to push the image closer to the boundary where the model says “tomato.” Machine learning has neither the context nor the common sense to understand that tomatoes don’t appear in mid-air. Could a delivery drone be subverted to become a weapon by causing it to misunderstand its surroundings? Almost certainly. Don’t dismiss these examples as academic. A stop sign with a few pixels changed in the lab may not be different from a stop sign that has been used for target practice during hunting season.

Impersonation attacks attempt to fool a model into misidentifying someone or something. The goal is frequently to gain unauthorized access to a system. For example, an attacker might want to trick a bank into misreading the amount written on a check. Fingerprints obtained from drinking glasses, or even high resolution photographs, can be used to fool fingerprint authentication. South Park trolled Alexa and Google Home users by using the words “Alexa” and “OK Google” repeatedly in an episode, triggering viewers’ devices; the devices weren’t able to distinguish between the show voices and real ones. The next generation of impersonation attacks will be “deep fake” videos that place words in the mouths of real people.

Inversion means using an API to gather information about a model, and using that information to attack it. Inversion can also mean using an API to obtain private information from a model, perhaps by retrieving data and de-anonymizing it. In “The Secret Sharer: Measuring Unintended Neural Network Memorization & Extracting Secrets,” the authors show that machine learning models tend to memorize all their training data, and that it’s possible to extract protected information from a model. Common approaches to protecting information don’t work; the model still incorporates secret information in ways that can be extracted. Differential privacy—the practice of carefully inserting extraneous data into a data set in ways that don’t change its statistical properties—has some promise, but with significant cost: the authors point out that training is much slower. Furthermore, the number of developers who understand and can implement differential privacy is small.

While this may sound like an academic concern, it’s not; writing a script to probe machine learning applications isn’t difficult. Furthermore, Michael Veale and others write that inversion attacks raise legal problems. Under the GDPR, if protected data is memorized by models, are those models subject to the same regulations as personal data? In that case, developers would have to remove personal data from models—not just the training data sets—on request; it would be very difficult to sell products that incorporated models, and even techniques like automated model generation could become problematic. Again, the authors point to differential privacy, but with the caution that few companies have the expertise to deploy models with differential privacy correctly.

Other vulnerabilities, other attacks

This brief taxonomy of vulnerabilities doesn’t come close to listing all the problems that machine learning will face in the field. Many of these vulnerabilities are easily exploited. You can probe Amazon to find out what products are recommended along with your products, possibly finding out who your real competitors are, and discovering who to attack. You might even be able to reverse-engineer how Amazon makes recommendations and use that knowledge to influence the recommendations they make.

More complex attacks have been seen in the field. One involves placing fake reviews on an Amazon seller’s site, so that when the seller removes the reviews, Amazon bans the seller for review manipulation. Is this an attack against machine learning? The attacker tricks the human victim into violating Amazon’s rules. Ultimately, though, it’s the machine learning system that’s tricked into taking an incorrect action (banning the victim) that it could have prevented.

“Google bowling” means creating large numbers of links to a competitor’s website in hopes that Google’s ranking algorithm will penalize the competitor for purchasing bulk links. It’s similar to the fake review attack, except that it doesn’t require a human intermediary; it’s a direct attack against the algorithm that analyzes inbound links.

Advertising was one of the earliest adopters of machine learning, and one of the earliest victims. Click fraud is out of control, and the machine learning community is reluctant to talk about (or is unaware of) the issue—even though, as online advertising becomes ever more dependent on machine learning, fraudsters will learn how to attack models directly in their attempts to appear legitimate. If click data is unreliable, then models built from that data are unreliable, along with any results or recommendations generated by those models. And click fraud is similar to many attacks against recommendation systems and trend analysis. Once a “fake news” item has been planted, it’s simple to make it trend with some automated clicks. At that point, the recommendation takes over, generating recommendations which in turn generate further clicks. Anything automated is prone to attack, and automation allows those attacks to take place at scale.

The advent of autonomous vehicles, ranging from cars to drones, presents yet another set of threats. If the machine learning systems on an autonomous vehicle are vulnerable to attack, a car or truck could conceivably be used as a murder weapon. So could a drone—either a weaponized military drone or a consumer drone. The military already knows that drones are vulnerable; in 2011, Iran captured a U.S. drone, possibly by spoofing GPS signals. We expect to see attacks on “smart” consumer health devices and professional medical devices, many of which we know are already vulnerable.

Taking action

Merely scolding and thinking about possible attacks won’t help. What can be done to defend machine learning models? First, we can start with traditional software. The biggest problem with insecure software isn’t that we don’t understand security; it’s that software vendors, and software users, never take the basic steps they would need to defend themselves. It’s easy to feel defenseless before hyper-intelligent hackers, but the reality is that sites like Equifax become victims because they didn’t take basic precautions, such as installing software updates. So, what do machine learning developers need to do?

Security audits are a good starting point. What are the assets that you need to protect? Where are they, and how vulnerable are they? Who has access to those resources, and who actually needs that access? How can you minimize access to critical data? For example, a shipping system needs customer addresses, but it doesn’t need credit card information; a payment system needs credit card information, but not complete purchase histories. Can this data be stored and managed in separate, isolated databases? Beyond that, are basic safeguards in place, such as two-factor authentication? It’s easy to fault Equifax for not updating their software, but almost any software system depends on hundreds, if not thousands, of external libraries. What strategy do you have in place to ensure they’re updated, and that updates don't break working systems?

Like conventional software, machine learning systems should use monitoring systems that generate alerts to notify staff when something abnormal or suspicious occurs. Some of these monitoring systems are already using machine learning for anomaly detection—which means the monitoring software itself can be attacked.

Penetration testing is a common practice in the online world: your security staff (or, better, consultants) attack your site to discover its vulnerabilities. Attack simulation is an extension of penetration testing that shows you “how attackers actually achieve goals against your organization.” What are they looking for? How do they get to it? Can you gain control over a system by poisoning its inputs?

Tools for testing computer vision systems by generating "adversarial images" are already appearing, such as cleverhans and IBM’s ART. We are starting to see papers describing adversarial attacks against speech recognition systems. Adversarial input is a special case of a more general problem. Most machine learning developers assume their training data is similar to the data their systems will face in the real world. That’s an idealized best case. It’s easy to build a face identification system if all your faces are well-lit, well-focused, and have light-skinned subjects. A working system needs to handle all kinds of images, including images that are blurry, badly focused, poorly lighted—and have dark-skinned subjects.

Safety verification is a new area for AI research, still in its infancy. Safety verification asks questions like whether models can deliver consistent results, or whether small changes in the input lead to large changes in the output. If machine learning is at all like conventional software, we expect an escalating struggle between attackers and defenders; better defenses will lead to more sophisticated attacks, which will lead to a new generation of defenses. It will never be possible to say that a model has been “verifiably safe.” But it is important to know that a model has been tested, and that it is reasonably well-behaved against all known attacks.

Model explainability has become an important area of research in machine learning. Understanding why a model makes specific decisions is important for several reasons, not the least of which is that it makes people more comfortable with using machine learning. That “comfort” can be deceptive, of course. But being able to ask models why they made particular decisions will conceivably make it easier to see when they’ve been compromised. During development, explainability will make it possible to test how easy it is for an adversary to manipulate a model, in applications from image classification to credit scoring. In addition to knowing what a model does, explainability will tell us why, and help us build models that are more robust, less subject to manipulation; understanding why a model makes decisions should help us understand its limitations and weaknesses. At the same time, it’s conceivable that explainability will make it easier to discover weaknesses and attack vectors. If you want to poison the data flowing into a model, it can only help to know how the model responds to data.

In “Deep Automation in Machine Learning,” we talked about the importance of data lineage and provenance, and tools for tracking them. Lineage and provenance are important whether or not you’re developing the model yourself. While there are many cloud platforms to automate model building and even deployment, ultimately your organization is responsible for the model’s behavior. The downside of that responsibility includes everything from degraded profits to legal liability. If you don’t know where your data is coming from and how it has been modified, you have no basis for knowing whether your data has been corrupted, either through accident or malice.

“Datasheets for Datasets” proposes a standard set of questions about a data set’s sources, how the data was collected, its biases, and other basic information. Given a specification that records a data set’s properties, it should be easy to test and detect sudden and unexpected changes. If an attacker corrupts your data, you should be able to detect that and correct it up front; if not up front, then later in an audit.

Datasheets are a good start, but they are only a beginning. Whatever tools we have for tracking data lineage and provenance need to be automated. There will be too many models and data sets to rely on manual tracking and audits.

Balancing openness against tipping off adversaries

In certain domains, users and regulators will increasingly prefer machine learning services and products that can provide simple explanations for how automated decisions and recommendations are being made. But we’ve already seen that too much information can lead to certain parties gaming models (as in SEO). How much to disclose depends on the specific application, domain, and jurisdiction.

This balancing act is starting to come up in machine learning and related areas that involve the work of researchers (who tend to work in the open) who are up against adversaries who prize unpublished vulnerabilities. The question of whether or not to “temporarily hold back” research results is a discussion that the digital media forensics community has been having. In a 2018

https://ift.tt/2Vrb0Ym

[repmywind02199]

You created a machine learning application. Now make sure it’s secure.

You created a machine learning application. Now make sure it’s secure.

The software industry has demonstrated, all too clearly, what happens when you don’t pay attention to security.

In a recent post, we described what it would take to build a sustainable machine learning practice. By “sustainable,” we mean projects that aren’t just proofs of concepts or experiments. A sustainable practice means projects that are integral to an organization’s mission: projects by which an organization lives or dies. These projects are built and supported by a stable team of engineers, and supported by a management team that understands what machine learning is, why it’s important, and what it’s capable of accomplishing. Finally, sustainable machine learning means that as many aspects of product development as possible are automated: not just building models, but cleaning data, building and managing data pipelines, testing, and much more. Machine learning will penetrate our organizations so deeply that it won’t be possible for humans to manage them unassisted.

Organizations throughout the world are waking up to the fact that security is essential to their software projects. Nobody wants to be the next Sony, the next Anthem, or the next Equifax. But while we know how to make traditional software more secure (even though we frequently don’t), machine learning presents a new set of problems. Any sustainable machine learning practice must address machine learning’s unique security issues. We didn’t do that for traditional software, and we’re paying the price now. Nobody wants to pay the price again. If we learn one thing from traditional software’s approach to security, it’s that we need to be ahead of the curve, not behind it. As Joanna Bryson writes, “Cyber security and AI are inseparable.”

The presence of machine learning in any organization won’t be a single application, a single model; it will be many applications, using many models—perhaps thousands of models, or tens of thousands, automatically generated and updated. Machine learning on low-power edge devices, ranging from phones to tiny sensors embedded in assembly lines, tools, appliances, and even furniture and building structures, increases the number of models that need to be monitored. And the advent of 5G mobile services, which significantly increases the network bandwidth to mobile devices, will make it much more attractive to put machine learning at the edge of the network. We anticipate billions of machines, each of which may be running dozens of models. At this scale, we can't assume that we can deal with security issues manually. We need tools to assist the humans responsible for security. We need to automate as much of the process as possible, but not too much, giving humans the final say.

In “Lessons learned turning machine learning models into real products and services,” David Talby writes that “the biggest mistake people make with regard to machine learning is thinking that the models are just like any other type of software.” Model development isn’t software development. Models are unique—the same model can’t be deployed twice; the accuracy of any model degrades as soon as it is put into production; and the gap between training data and live data, representing real users and their actions, is huge. In many respects, the task of modeling doesn’t get started until the model hits production, and starts to encounter real-world data.

Unfortunately, one characteristic that software development has in common with machine learning is a lack of attention to security. Security tends to be a low priority. It gets some lip service, but falls out of the picture when deadlines get tight. In software, that’s been institutionalized in the “move fast and break things” mindset. If you’re building fast, you’re not going to take the time to write sanitary code, let alone think about attack vectors. You might not “break things,” but you’re willing to build broken things; the benefits of delivering insecure products on time outweigh the downsides, as Daniel Miessler has written. You might be lucky; the vulnerabilities you create may never be discovered. But if security experts aren’t part of the development team from the beginning, if security is something to be added on at the last minute, you’re relying on luck, and that’s not a good position to be in. Machine learning is no different, except that the pressure of delivering a product on time is even greater, the issues aren’t as well understood, the attack surface is larger, the targets are more valuable, and companies building machine learning products haven’t yet engaged with the problems.

What kinds of attacks will machine learning systems see, and what will they have to defend against? All of the attacks we have been struggling with for years, but there are a number of vulnerabilities that are specific to machine learning. Here’s a brief taxonomy of attacks against machine learning:

Poisoning, or injecting bad (“adversarial”) data into the training data. We’ve seen this many times in the wild. Microsoft’s Tay was an experimental chatbot that was quickly taught to spout racist and anti-semitic messages by the people who were chatting with it. By inserting racist content into the data stream, they effectively gained control over Tay’s behavior. The appearance of “fake news” in channels like YouTube, Facebook, Twitter, and even Google searches, was similar: once fake news was posted, users were attracted to it like flies, and the algorithms that made recommendations “learned” to recommend that content. danah boyd has argued that these incidents need to be treated as security issues, intentional and malicious corruption of the data feeding the application, not as isolated pranks or algorithmic errors.

Any machine learning system that constantly trains itself is vulnerable to poisoning. Such applications could range from customer service chat bots (can you imagine a call center bot behaving like Tay?) to recommendation engines (real estate redlining might be a consequence) or even to medical diagnosis (modifying recommended drug dosages). To defend against poisoning, you need strong control over the training data. Such control is difficult (if not impossible) to achieve. “Black hat SEO” to improve search engine rankings is nothing if not an early (and still very present) example of poisoning. Google can’t control the incoming data, which is everything that is on the web. Their only recourse is to tweak their search algorithms constantly and penalize abusers for their behavior. In the same vein, bots and troll armies have manipulated social media feeds to spread views ranging from opposition to vaccination to neo-naziism.

Evasion, or crafting input that causes a machine learning system to misclassify it. Again, we’ve seen this both in the wild and in the lab. CV Dazzle uses makeup and hair styles as “camouflage against face recognition technology.” Other research projects have shown that it’s possible to defeat image classification by changing a single pixel in an image: a ship becomes a car, a horse becomes a frog. Or, just as with humans, image classifiers can miss an unexpected object that’s out of context: an elephant in the room, for example. It’s a mistake to think that computer vision systems “understand” what they see in ways that are similar to humans. They’re not aware of context, they don’t have expectations about what’s normal; they’re simply doing high-stakes pattern matching. Researchers have reported similar vulnerabilities in natural language processing, where changing a word, or even a letter, in a way that wouldn’t confuse human researchers causes machine learning to misunderstand a phrase.

Although these examples are often amusing, it’s worth thinking about real-world consequences: could someone use these tricks to manipulate the behavior of autonomous vehicles? Here’s how that could work: I put a mark on a stop sign—perhaps by sticking a fragment of a green sticky note at the top. Does that make an autonomous vehicle think the stop sign is a flying tomato, and if so, would the car stop? The alteration doesn’t have to make the sign “look like” a tomato to a human observer; it just has to push the image closer to the boundary where the model says “tomato.” Machine learning has neither the context nor the common sense to understand that tomatoes don’t appear in mid-air. Could a delivery drone be subverted to become a weapon by causing it to misunderstand its surroundings? Almost certainly. Don’t dismiss these examples as academic. A stop sign with a few pixels changed in the lab may not be different from a stop sign that has been used for target practice during hunting season.

Impersonation attacks attempt to fool a model into misidentifying someone or something. The goal is frequently to gain unauthorized access to a system. For example, an attacker might want to trick a bank into misreading the amount written on a check. Fingerprints obtained from drinking glasses, or even high resolution photographs, can be used to fool fingerprint authentication. South Park trolled Alexa and Google Home users by using the words “Alexa” and “OK Google” repeatedly in an episode, triggering viewers’ devices; the devices weren’t able to distinguish between the show voices and real ones. The next generation of impersonation attacks will be “deep fake” videos that place words in the mouths of real people.

Inversion means using an API to gather information about a model, and using that information to attack it. Inversion can also mean using an API to obtain private information from a model, perhaps by retrieving data and de-anonymizing it. In “The Secret Sharer: Measuring Unintended Neural Network Memorization & Extracting Secrets,” the authors show that machine learning models tend to memorize all their training data, and that it’s possible to extract protected information from a model. Common approaches to protecting information don’t work; the model still incorporates secret information in ways that can be extracted. Differential privacy—the practice of carefully inserting extraneous data into a data set in ways that don’t change its statistical properties—has some promise, but with significant cost: the authors point out that training is much slower. Furthermore, the number of developers who understand and can implement differential privacy is small.

While this may sound like an academic concern, it’s not; writing a script to probe machine learning applications isn’t difficult. Furthermore, Michael Veale and others write that inversion attacks raise legal problems. Under the GDPR, if protected data is memorized by models, are those models subject to the same regulations as personal data? In that case, developers would have to remove personal data from models—not just the training data sets—on request; it would be very difficult to sell products that incorporated models, and even techniques like automated model generation could become problematic. Again, the authors point to differential privacy, but with the caution that few companies have the expertise to deploy models with differential privacy correctly.

Other vulnerabilities, other attacks

This brief taxonomy of vulnerabilities doesn’t come close to listing all the problems that machine learning will face in the field. Many of these vulnerabilities are easily exploited. You can probe Amazon to find out what products are recommended along with your products, possibly finding out who your real competitors are, and discovering who to attack. You might even be able to reverse-engineer how Amazon makes recommendations and use that knowledge to influence the recommendations they make.

More complex attacks have been seen in the field. One involves placing fake reviews on an Amazon seller’s site, so that when the seller removes the reviews, Amazon bans the seller for review manipulation. Is this an attack against machine learning? The attacker tricks the human victim into violating Amazon’s rules. Ultimately, though, it’s the machine learning system that’s tricked into taking an incorrect action (banning the victim) that it could have prevented.

“Google bowling” means creating large numbers of links to a competitor’s website in hopes that Google’s ranking algorithm will penalize the competitor for purchasing bulk links. It’s similar to the fake review attack, except that it doesn’t require a human intermediary; it’s a direct attack against the algorithm that analyzes inbound links.

Advertising was one of the earliest adopters of machine learning, and one of the earliest victims. Click fraud is out of control, and the machine learning community is reluctant to talk about (or is unaware of) the issue—even though, as online advertising becomes ever more dependent on machine learning, fraudsters will learn how to attack models directly in their attempts to appear legitimate. If click data is unreliable, then models built from that data are unreliable, along with any results or recommendations generated by those models. And click fraud is similar to many attacks against recommendation systems and trend analysis. Once a “fake news” item has been planted, it’s simple to make it trend with some automated clicks. At that point, the recommendation takes over, generating recommendations which in turn generate further clicks. Anything automated is prone to attack, and automation allows those attacks to take place at scale.

The advent of autonomous vehicles, ranging from cars to drones, presents yet another set of threats. If the machine learning systems on an autonomous vehicle are vulnerable to attack, a car or truck could conceivably be used as a murder weapon. So could a drone—either a weaponized military drone or a consumer drone. The military already knows that drones are vulnerable; in 2011, Iran captured a U.S. drone, possibly by spoofing GPS signals. We expect to see attacks on “smart” consumer health devices and professional medical devices, many of which we know are already vulnerable.

Taking action

Merely scolding and thinking about possible attacks won’t help. What can be done to defend machine learning models? First, we can start with traditional software. The biggest problem with insecure software isn’t that we don’t understand security; it’s that software vendors, and software users, never take the basic steps they would need to defend themselves. It’s easy to feel defenseless before hyper-intelligent hackers, but the reality is that sites like Equifax become victims because they didn’t take basic precautions, such as installing software updates. So, what do machine learning developers need to do?

Security audits are a good starting point. What are the assets that you need to protect? Where are they, and how vulnerable are they? Who has access to those resources, and who actually needs that access? How can you minimize access to critical data? For example, a shipping system needs customer addresses, but it doesn’t need credit card information; a payment system needs credit card information, but not complete purchase histories. Can this data be stored and managed in separate, isolated databases? Beyond that, are basic safeguards in place, such as two-factor authentication? It’s easy to fault Equifax for not updating their software, but almost any software system depends on hundreds, if not thousands, of external libraries. What strategy do you have in place to ensure they’re updated, and that updates don't break working systems?

Like conventional software, machine learning systems should use monitoring systems that generate alerts to notify staff when something abnormal or suspicious occurs. Some of these monitoring systems are already using machine learning for anomaly detection—which means the monitoring software itself can be attacked.

Penetration testing is a common practice in the online world: your security staff (or, better, consultants) attack your site to discover its vulnerabilities. Attack simulation is an extension of penetration testing that shows you “how attackers actually achieve goals against your organization.” What are they looking for? How do they get to it? Can you gain control over a system by poisoning its inputs?

Tools for testing computer vision systems by generating "adversarial images" are already appearing, such as cleverhans and IBM’s ART. We are starting to see papers describing adversarial attacks against speech recognition systems. Adversarial input is a special case of a more general problem. Most machine learning developers assume their training data is similar to the data their systems will face in the real world. That’s an idealized best case. It’s easy to build a face identification system if all your faces are well-lit, well-focused, and have light-skinned subjects. A working system needs to handle all kinds of images, including images that are blurry, badly focused, poorly lighted—and have dark-skinned subjects.

Safety verification is a new area for AI research, still in its infancy. Safety verification asks questions like whether models can deliver consistent results, or whether small changes in the input lead to large changes in the output. If machine learning is at all like conventional software, we expect an escalating struggle between attackers and defenders; better defenses will lead to more sophisticated attacks, which will lead to a new generation of defenses. It will never be possible to say that a model has been “verifiably safe.” But it is important to know that a model has been tested, and that it is reasonably well-behaved against all known attacks.

Model explainability has become an important area of research in machine learning. Understanding why a model makes specific decisions is important for several reasons, not the least of which is that it makes people more comfortable with using machine learning. That “comfort” can be deceptive, of course. But being able to ask models why they made particular decisions will conceivably make it easier to see when they’ve been compromised. During development, explainability will make it possible to test how easy it is for an adversary to manipulate a model, in applications from image classification to credit scoring. In addition to knowing what a model does, explainability will tell us why, and help us build models that are more robust, less subject to manipulation; understanding why a model makes decisions should help us understand its limitations and weaknesses. At the same time, it’s conceivable that explainability will make it easier to discover weaknesses and attack vectors. If you want to poison the data flowing into a model, it can only help to know how the model responds to data.

In “Deep Automation in Machine Learning,” we talked about the importance of data lineage and provenance, and tools for tracking them. Lineage and provenance are important whether or not you’re developing the model yourself. While there are many cloud platforms to automate model building and even deployment, ultimately your organization is responsible for the model’s behavior. The downside of that responsibility includes everything from degraded profits to legal liability. If you don’t know where your data is coming from and how it has been modified, you have no basis for knowing whether your data has been corrupted, either through accident or malice.

“Datasheets for Datasets” proposes a standard set of questions about a data set’s sources, how the data was collected, its biases, and other basic information. Given a specification that records a data set’s properties, it should be easy to test and detect sudden and unexpected changes. If an attacker corrupts your data, you should be able to detect that and correct it up front; if not up front, then later in an audit.

Datasheets are a good start, but they are only a beginning. Whatever tools we have for tracking data lineage and provenance need to be automated. There will be too many models and data sets to rely on manual tracking and audits.

Balancing openness against tipping off adversaries

In certain domains, users and regulators will increasingly prefer machine learning services and products that can provide simple explanations for how automated decisions and recommendations are being made. But we’ve already seen that too much information can lead to certain parties gaming models (as in SEO). How much to disclose depends on the specific application, domain, and jurisdiction.

This balancing act is starting to come up in machine learning and related areas that involve the work of researchers (who tend to work in the open) who are up against adversaries who prize unpublished vulnerabilities. The question of whether or not to “temporarily hold back” research results is a discussion that the digital media forensics community has been having. In a 2018

https://ift.tt/2Vrb0Ym

[brilliantsprout-blog]

10 Beginners Machine http://bit.ly/2Vnzoem

10 Beginners Machine Learning Tutorials Machine Learning is one of the fastest developing fields that provide excellent job opportunities. So if you are looking for an open niche in which you will be able to build a successful career and be proud of what you do you should start studying Machine Learning. There are a lot of Machine Learning tutorials for beginners available online and you simply need to choose the most suitable one. Check the following top 10 courses which will help you to acquire knowledge in the most effective way. What Are The Prerequisites To Learn Machine Learning? To take Machine Learning courses you dont need to be a Master in Computer Science. However you should have at least some basic knowledge related to this field. Most courses are available for students with the following skills: At least an intermediate level of exposure to calculus linear algebra and probability. Familiarity with at least one programming language preferably Python. Top 10 Machine Learning Tutorials for Beginners Machine Learning (Stanford) Machine Learning course developed by Andrew Ng is popular among the students and for a good reason. It provides beginners with a comprehensive overview of all important topics starting from parametric/non-parametric algorithms and neural networks and ending with clustering and deep learning. However the best advantage of Stanfords course is its practical part. There are a lot of real-life case studies and assignments which helps to understand how Machine Learning works in real life. At the end of the course online students get verified certificates. Course Highlights Duration 55 hours (7 hours per week) Course completion certificate Course available in 6 languages If you dont need a certificate you can study for free. However some assignments will not be available. Applied Machine Learning Course (Columbia Engineering) This course consists of two parts: supervised and unsupervised learning. Each part includes 5 modules 10 modules in total. If you like a traditional way of studying you will definitely like this course. You will get access to faculty video lectures and moderated discussion boards. Even more you will be able to communicate with real teachers online. They will guide you through the studying process pointing out your mistakes and answering all your questions. Course Highlights Course fees $1200 Duration 3 months (6-8 hours per week) Verified digital certificate from EMERITUS in collaboration with Columbia Engineering Executive Education. Machine Learning: Level 1 and Level 2 (Kaggle) If you are not sure whether machine learning is your cup of tea and don't want to spend a lot on tutorials yet you can take free courses offered by Kaggle. This course consists of two levels. Level 1 includes 6 lectures and 6 practical exercises. It will help you to understand how to apply obtained knowledge to real-life situations. Level 2 will help you to master new skills and learn more about handling missing values data leakage and application of partial dependents plots. Course Highlights Course is free High level of Python skills is required to take this course No certificate is given for this course. Machine Learning Crash Course (Google) Google designed outstanding tutorials in partnership with Kaggle. The short course allows studying fundamental machine learning concepts and getting real-world experience. Besides if you want to learn more about Artificial Intelligence you can visit the Google AI Education page and get access to a huge up-to-date library with free training resources. Course Highlights Course is free Duration 25 hours Self-paced course format. Weka Machine Learning Mini-Course If you have a goal to learn basic algorithm of Machine Learning in a short period of time you can try this free mini-course. In a few lessons you will learn how to use Weka software. At the end of the course you will be able to estimate model performance and write classification regression and ensemble algorithms. However keep in mind that you will learn how to work with the digits only. If you want to work with the text you should better find other tutorials. Course Highlights Course is free Installation of Weka software is needed No certificate is given for this course. Data Science and Machine Learning Bootcamp with R If you are not familiar with Python you can choose another way to master Machine Learning skills. You can study R programming and learn how to apply it for data visualization and manipulation. If you take this course on Udemy you will become an expert in using R for Machine Learning algorithms. Course Highlights Course fees $194.99 (Sometimes discounts up to 90% are available) 175 hours of video lectures 8 articles 3 recourses to download are included Course completion certificate. Essential Math for Machine Learning: R Edition If you lack math skills you will find it challenging to study Machine Learning. Its highly recommended to take specialized Essential Math for Machine Learning course to ensure that you will not face any difficulties in mastering new skills at the next stages. Mathew Millington CEO of Studicus.com says: Its crucially important to learn fundamentals first. It will help you to achieve great results in the future. So if you feel that you need to prepare yourself before jumping into Machine Learning feel free to take additional tutorials for beginners. Course Highlights Self-paced course format Duration 6 weeks (6-8 hours per week) You can get a course completion certificate if you pay $99 (if you dont need a certificate you can study for free) Get Quick Overview of Machine Learning Principles If you have no time to study but urgently need to know what Machine Learning is all about you can visit Pluralsigh to access short video tutorials. Scientists developers professors and other experts in the field offer short explanatory videos. These video materials will help to learn fundamentals in a hassle-free way. You can check the following popular courses: Getting Started with Azure Machine Learning Understanding Machine Learning with Python How to Think About Machine Learning Algorithms In Conclusion There are a lot of tutorials available online and you can choose that one which meets your preferences and your budget. However you should always consider expenses on education as a significant investment in your future. The higher price (in the form of your money time and efforts) you pay the higher returns (in the form of higher salary and better benefits) you will receive. Linda Grandes is a full-time blogger at Studyton.com and a true expert in writing. She is interested in a variety of topics starting from education and ending at modern art. Thats why the idea of her blog cannot be narrowed to one topic as well. Linda also has vast experience in editing texts and knows all the angles of academic and professional writing thanks to her productive cooperation with WoWGrade.com. Linda is a passionate traveler and she is eager to learn new things and meet new people.

[crookedtreepoetry]

Introduction To data Science With Python

Fortunately, learning Python and other programming fundamentals is as attainable as ever. Exploratory data analysis is an method used to analyze massive knowledge sets to summarize their main traits. This course of uses visible strategies to derive priceless insights. I consent to permit Cognitive Class to use cookies to seize product utilization analytics. The data from these cookies will only be used for product utilization on Cognitive Class domains, and this usage data will not be shared outside of Cognitive Class. The product usage will be used for business reporting and product usage understanding.

Additionally, you may also comply with one of the best programs on Machine Learning course from Yaser Abu-Mostafa. If you want more lucid rationalization for the techniques, you presumably can go for the Machine learning course from Andrew Ng and observe the workout routines on Python. You should also check out the ‘Introduction to Data Science‘ course to offer yourself an enormous enhance in your quest to land an information scientist role. Now that you have made up your mind, it's time to arrange your machine. The easiest approach to proceed is to only obtain Anaconda from Continuum.io .

Data science specialists anticipate this pattern to proceed with growing improvement within the Python ecosystem. And whereas your journey to be taught Python programming could also be simply beginning, it’s nice to know that employment alternatives are ample as well. Let’s now perceive how the confusion matrix decides the accuracy of the model.

You ought to begin by understanding the basics of the language, libraries and knowledge structure. Rather than studying opinions, take a glance at this more objective article about how Python and R handle related knowledge science tasks, and see which one seems more approachable to you. There are plenty of free Python for knowledge science tutorials out there. If you don't need to pay to learn Python, these could be a good possibility — and the link in the previous sentence consists of dozens, separated out by difficulty level and focus space. This is because Python can also be utilized in quite lots of other programming disciplines from recreation growth to cell apps. Generic "be taught Python" assets try to educate a little bit of every thing, however this means you'll be learning fairly a few issues that aren't really related to data science work.

We will use the Logistic Regression algorithm to construct the model. Logistic Regression is suitable when the dependent variable is binary. It is obligatory to obtain user consent prior to operating these cookies on your web site.

Data visualization is the graphical representation of knowledge in order to interactively and effectively convey insights to purchasers, prospects, and stakeholders normally. It is a way to summarize your findings and show it in a kind that facilitates interpretation and may help in figuring out patterns or tendencies. In this course you will discover ways to create interesting graphics and charts and customize them to make them more practical and extra pleasing to your viewers. Includes subtopics similar to pure language processing, and image manipulation with libraries such as OpenCV.

It comes packaged with most of the issues you'll need ever. The major downside of taking this route is that you'll need to attend for Continuum to update their packages, even when there may be an replace available to the underlying libraries. Using Python and the pandas and matplotlib libraries, you begin analyzing, exploring, and visualizing the data. During this time, you’ll wish to make sure you’re cultivating these delicate expertise required to work with others, making sure you really understand the inner workings of the instruments you’re utilizing. Machine Learning Project — If you aspire to work as a data scientist, you undoubtedly will need a project that reveals off your ML chops .

Python is a a lot better language for all-around work, meaning that your Python skills can be extra transferrable to different disciplines. It's also barely extra in style, and a few would argue that it is the simpler of the two to learn . (Of course, you may additionally have to study some SQL regardless of which of Python or R you decide to be your primary programming language). Really, it all is dependent upon your desired timeline, free time you could dedicate to learn Python programming and the tempo at which you learn.

It's possible to work as an information scientist utilizing both Python or R. Each language has its strengths and weaknesses, and both are widely-used in the trade. Python is more in style overall, but R dominates in some industries .

At the rate that demand is rising, there are exponential opportunities to learn. Continue studying, collaborating, and conversing with others, and you’re certain to maintain interest and a competitive edge over time. Matplotlib — A visualization library that makes it fast and easy to generate charts out of your knowledge. You can attempt programming issues like calculators for an internet game, or a program that fetches the weather from Google in your city. You also can construct simple games and apps to help you familiarize yourself with working with Python.

Python as a programming language has become very fashionable in recent occasions. It has been utilized in data science, IoT, AI, and other applied sciences, which has added to its popularity. This beginner-friendly Python course will take you from zero to programming in Python in a matter of hours. You'll have the ability to write your own Python scripts and carry out basic hands-on information analysis using our Jupyter-based lab surroundings. This introduction to Python will kickstart your studying ofPython for information science, in addition to programming generally.

Explore more on - Data Science Institutes in Hyderabad

360DigiTMG - Data Analytics, Data Science Course Training Hyderabad

 Address:-2-56/2/19, 3rd floor, Vijaya towers, near Meridian school, Ayyappa Society Rd, Madhapur, Hyderabad, Telangana 500081

Contact us ( 099899 94319 )

https://360digitmg.com/data-science-course-training-in-hyderabad

Hours: Sunday - Saturday 7 AM - 11 PM

[ardas-group]

How to Make a Chatbot in Messenger: Main Benefits for Your Business

The classic scheme of communication of any business with a person has always been through the phone. If you needed a plumber, you called the service and ordered, or if you wanted pizza at three in the morning, you called too. The telephone was the main channel of communication and service to people. Further, with the advent of mobile phones, it was transformed into applications. First, there were help desks available through the browser, then there were smartphones with an endless number of applications that we download daily.

A huge number of applications on the phone lead to an inconvenience of use since for each separate restaurant or cinema, you need to open a separate application to contact the operator. This is where chatbots in messengers come to the rescue, which solves the issue of centralization, and now in the same application where you communicate with your loved ones, you can book a table for the evening in your favorite restaurant. What you need to create a chatbot in a messenger application and what are the benefits for your SaaS project, we will consider in this article.

8 Benefits of Chatbots for Businesses

With the advent of chatbots, there have been many changes in the business. Companies and customers are interested in this tool, but not always keep up with the updates in the industry. Let's look at why brands are implementing chatbots, what they are guided by when creating them, and how a chatbot can help a business.

Link openness is much higher than email.

Faster communication with the client (everyone is now hanging on social networks).

You definitely won't end up in spam.

People are pissed off by the red dots of an unread message, and they often open messenger to watch.

The chatbot can immediately reply to messages without delay in opening mail. Therefore, the client does not have time to "cool down" - efficiency is now winning.

During webinars, there are technical issues that are quickly resolved through the messenger.

Bot creates the illusion of personal communication with clients.

You can quickly pick up your chat history.

How Does a Chatbot Work?

80% of companies in the world have created chatbots or plan to use them in 2020, follows from data from Oracle, a leading American corporation, as it's the world's largest software developer. According to Facebook, from 2017 to 2018, the activity of correspondence using chatbots increased more than five times. The reason for the popularity of chatbots is obvious - they keep the user engaged, provide more reach, and reduce the amount of routine work dramatically.

Many experts say: "Make a chatbot - you will sell." But it's not that simple. A chatbot can sell, but if an unprepared person immediately clicks on your Facebook ad button, gets into the messenger and the chatbot writes to him: “Buy my socks. This is the price,” - this person will probably not buy anything.

The reason is not that he does not need the product - he went through an advertisement. It's just that the user is "undercooked" enough to buy from your chatbot.

What can a chatbot in messenger really do for SaaS?

Greet the customer;

Pull up answers by keywords;

Delayed mailing;

Answer questions, help choose programs, products;

Segment your audience;

Count discounts, promotions, test results, and so on, depending on the choice of the answer;

Sell to the person's preferences.

Building a Chatbot from Scratch

This is the landscape of the chatbot industry, created by the American service KeyReply. It reflects the key market players and tools for developing bots.

For comparison, we will consider several options for instant messengers in which you can create chatbots. For example, Viber chatbot technology:

The client writes a message to the chat, which goes to the Viber server;

Viber accesses your server via an HTTP request;

Your server provides an accurate response to this, which goes back to the Viber server;

The answer comes to the chat.

Why is it necessary to understand this scheme? Firstly, many clients think that when they create a chatbot in a messenger, it will immediately become super-smart and will answer any questions. This part depends entirely on your server. The logic of answers is built from the client's side and it depends only on you how many questions the chatbot can answer accurately. In this scheme, any messenger is just an intermediary.

The Step-by-step Guide to Creating a Chatbot

To create a bot, follow these steps:

You need to have an active account in the messenger;

Register with the local Admin Panel where chatbots are created;

Fill out a questionnaire about the future bot. Usually, the questionnaire contains such fields as:

Account Image - community logo;

Account Name - the name that will be displayed in the header of the bot;

URI - by which the bot can be found.

Obtaining a token to work with the messenger server, it will be used to authenticate the request to the messenger API.

Setting up a web server and connecting an SSL certificate;

Installing webhook, this action is performed once; Webhook reports the following data to the Viber server:

A server address and endpoint that will receive messages from clients;

A list of events that the server will listen to;

Token;

Verification and publication of the chatbot.

What tricks you might face with chatbots for business

Сreating a chatbot in a messenger application, you need to understand the features of each and choose the one specifically that does not limit your possibilities as much as possible.

Be prepared for messenger platform limitations

Please note that each messenger has a number of technical limitations and if you are not ready for them, then you should seriously consider another chatbot option. For example, a ban from WhatsApp can be obtained for:

Sending a large number of messages at the same time. This is one of the main reasons when 100+ of the same messages fly from the number at the same time - WhatsApp algorithms calculate this easily.

Negative user reaction. If you are spamming, then 3-5 clicks on the "Complain" button are enough to block your number.

The client will work on any device with an internet connection and Docker support. For example, simple testing can be performed on a regular laptop. The minimum server requirements for the production environment are 250 GB SSD, 16 GB RAM, Quad-core CPU.

In general, the database should be running on a separate physical server (not where the Coreapp and Webapp containers are located). The admissible delay in data transmission between this server and the computer is no more than a few milliseconds.

Viber has its own list of restrictions that you should pay attention to:

Messages can be sent to all users who are subscribed (the user subscribes when he sends the bot the first message)

The maximum size of a JSON response is 30kb

The bot can send the user up to 100 messages per hour, the message counter is reset after the user replies to the message

The ability to send bulk messages opens only upon application and approval from Viber account managers.

The maximum length of the list of recipients of a mass message is 300 people, each such request can be executed up to 500 times in 10 seconds

If the user fails to deliver the message, Viber will attempt to resend this message for 14 days.

How can you customize your chatbot for SaaS?

Information management

There is no need to count on the fact that you will have a chatbot that will understand any of your text, written as you like. Most chatbots are not equipped with such cool intelligence. The overwhelming majority is somewhat reminiscent of USSD menus in SMS only in an improved version.

To control the bot, you, as the bot creator, must assume some kind of scenario. The user needs to send a command to get the desired result from the bot. When the bot receives the command, it will process it and react according to the scenario you have set. But imagine that the user wrote a command with an error, the bot, in this case, should warn that the command is not recognized and ask to try again. This is not entirely convenient for the user, and he can go looking for another bot with similar capabilities. You can lose a potential client, that's not good.

In this case, keyboards will help us. It just needs to be displayed to the user after the start of the dialogue with the bot, and when he clicks on the button he needs, the correct command you put in will be sent to the bot. In this case, we avoid possible mistakes, the bot interface looks more welcoming and professional. The quality of the bot is improved.

With each message, the bot can send a different set of buttons, both in terms of quantity and purpose. Buttons can be positioned directly below the message (InlineKeyboardButton) or docked below the text box for entering a message (KeyboardButton).

Receiving the information

In addition to text messages, bots can communicate using:

Images;

Audio files;

Video files;

Lists;

Carousels;

Buttons;

Receipts (for example, for a purchase);

Templates (often used when booking something, for example, air tickets);

Navigation menu.

Users can reply with text, emoticons, GIFs, images, audio, video, and pull chatbots into group chats.

What if the chatbot doesn't know how to answer the question?

Developers build a conversation map and, depending on the question, the person is given an automatic answer on websites, messengers, or applications. A chatbot is most convenient to use where processes are properly structured. For example, to buy a train ticket. Since the possible routes, schedule, availability of seats are known, the dialogue map is easy to build, and system errors are reduced to zero.

When the webserver runs out of answers to the customer's questions, the webserver can connect the operator to the chat. Each question of the client and the answer of the operator, the web server can save in the database for optimizing the bot.

The chatbot can answer questions based on the user's correspondence history or respond with prepared response templates.

Each new user will have an empty correspondence at startup, at which time the chatbot can send only one greeting message. When the user sends their first message, they automatically subscribe to the bot. Each message can contain an arbitrary keyboard with options for actions. Since each user has a unique ID, the web server can rely on the saved correspondence history of a particular user and, based on the history, send different responses to the client, or help the operator to join the conversation without the user noticing.

How to create a smart chatbot?

All messages entered by the user are transmitted and processed on the webserver. The web server can return an arbitrary keyboard with buttons, each button can perform some action, for example: go to the site, create a new message, write the user's mobile phone, write the user's location.

The simplest chatbot should have its own knowledge base, which is a set of possible user questions and their corresponding answers. The chatbot can select responses in several ways:

By keywords;

By phrase coincidence;

By a coincidence of context.

A kind of mini-problem for such simple bots is the identification of word forms and synonyms.

Dialogflow

Writing AI for a bot is not an easy task, so you can use a solution from Google. The following functions are available:

Creating answers to standard questions;

Creation of topics of conversation for non-standard or specific requests.

During the conversation, a suitable dialogue will be selected;

Training (adjusting bot responses based on the chat history);

Analytics (number of requests, the course of the dialogue, which topics are most often triggered).

Supported languages: English, Russian

Integration: Viber (deprecated, they suggest using the library for Nodejs), Telegram, Skype (deprecated), Twilio (deprecated), Facebook, Slack.

Aimylogic

Excellent service for creating all the familiar "text" bots and unusual bots for calling (for a paid plan). Fun but difficult to learn features like intents and entities. Suitable for creating AI and taking over the world.

It lacks the usual functions: a basket, creating an inline keyboard, sending a file, but you can create a bot that will guess what they write about.

Supported languages: English, Russian

Integration: WhatsApp, Facebook, VK.

Final thoughts

WhatsApp, Viber, Telegram, WeChat, Line, Facebook Messenger, and others. These messengers are convenient and they have become a familiar means of communication. Every day they fill our daily life more and more. We communicate with friends and relatives, colleagues in chats. And it became commonplace.

Thanks to the digitalization of life, all our affairs, events, entertainment apps are in our favorite smartphone and messengers. And why?

Because 60% of users who want to place an order for a product, product, or service prefer to write a message rather than call. Of these, 30% use instant messaging. It is not surprising that instant messengers have confidently begun to overtake social networks.

First, we stopped calling and switched to text communication, now we stop writing manually, and with one click we select from the suggested answers by the chatbot.

The main thing is to comply with the restrictions and try to create a chatbot as smart as possible. If you still have questions on chatbot development, our team will gladly take this headache away from you.

Originally taken from https://ardas-it.com/how-to-make-a-chatbot-in-messenger-main-benefits-for-your-business