#when algorithms didn’t control your content consumption and creation | Explore Tumblr posts and blogs

theinvinciblenoob · 7 years ago

Link

Jarno M. Koponen Contributor

Jarno M. Koponen is working on intelligent systems and human-centered personalization. He currently is product lead at Yle, one of the leading media houses in the Nordics.

More posts by this contributor

AI on your lock screen

The next AI is no AI

To put it mildly, news media has been on the sidelines in AI development. As a consequence, in the age of AI-powered personalized interfaces, the news organizations don’t anymore get to define what’s real news, or, even more importantly, what’s truthful or trustworthy. Today, social media platforms, search engines and content aggregators control user flows to the media content and affect directly what kind of news content is created. As a result, the future of news media isn’t anymore in its own hands. Case closed?

The (Death) Valley of news digitalization

There’s a history: News media hasn’t been quick or innovative enough to become a change maker in the digital world. Historically, news used to be the signal that attracted and guided people (and advertisers) in its own right. The internet and the exponential explosion of available information online changed that for good.

In the early internet, the portals channeled people to the content in which they were interested. Remember Yahoo? As the amount of information increased, the search engine(s) took over, changing the way people found relevant information and news content online. As the mobile technologies and interfaces started to get more prominent, social media with News Feed and tweets took over, changing again the way people discovered media content, now emphasizing the role of our social networks.

Significantly, news media didn’t play an active role in any of these key developments. Quite the opposite, it was late in utilizing the rise of the internet, search engines, content aggregators, mobile experience, social media and other new digital solutions to its own benefit.

The ad business followed suit. First news organizations let Google handle searches on their websites and the upcoming search champion got a unique chance to index media content. With the rise of social media, news organizations, especially in the U.S., turned to Facebook and Twitter to break the news rather than focusing on their own breaking news features. As a consequence, news media lost its core business to the rising giants of the new digital economy.

To put it very strongly, news media hasn’t ever been fully digital in its approach to user experience, business logic or content creation. Think paywalls and e-newspapers for the iPad! The internet and digitalization forced the news media to change, but the change was reactive, not proactive. The old, partly obsolete, paradigms of content creation, audience understanding, user experience and content distribution still actively affect the way news content is created and distributed today (and to be 110 percent clear — this is not about the storytelling and the unbelievable creativity and hard work done by ingenious journalists all around the globe).

Due to these developments, today’s algorithmic gatekeepers like Google and Facebook dominate the information flows and the ad business previously dominated by the news media. Significantly, personalization and the ad-driven business logic of today’s internet behemoths isn’t designed to let the news media flourish on its own terms ever again.

From observers to change makers

News media have been reporting the rise of the new algorithmic world order as an outside observer. And the reporting has been thorough, veracious and enlightening — the stories told by the news media have had a concrete effect on how people perceive our continuously evolving digital realities.

However, as the information flows have moved into the algorithmic black boxes controlled by the internet giants, it has become obvious that it’s very difficult or close to impossible for an outside observer to understand the dynamics that affect how or why a certain piece of information becomes newsworthy and widely spread. For the mainstream news media, Trump’s rise to the presidency came as a “surprise,” and this is but one example of the new dynamics of today’s digital reality.

And here’s a paradox. As the information moves closer to us, to the mobile lock screen and other surfaces that are available and accessible for us all the time, its origins and background motives become more ambiguous than ever.

The current course won’t be changed by commenting on or criticizing the actions of the ruling algorithmic platforms.

The social media combined with self-realizing feedback loops utilizing the latest machine learning methods, simultaneously being vulnerable for malicious or unintended gaming, has led us to the world of “alternative facts” and fake news. In this era of automated troll-hordes and algorithmic manipulation, the ideals of news media sound vitally important and relevant: Distribution of truthful and relevant information; nurturing the freedom of speech; giving the voice to the unheard; widening and enriching people’s worldview; supporting democracy.

But, the driving values of news media won’t ever be fully realized in the algorithmic reality if the news media itself isn’t actively developing solutions that shape the algorithmic reality.

The current course won’t be changed by commenting on or criticizing the actions of the ruling algorithmic platforms. #ChangeFacebook is not on the table for news media. New AI-powered Google News is controlled and developed by Google, based on its company culture and values, and thus can’t be directly affected by the news organizations.

After the rise of the internet and today’s algorithmic rule, we are again on the verge of a significant paradigm shift. Machine learning-powered AI solutions will have an increasingly significant impact on our digital and physical realities. This is again a time to affect the power balance, to affect the direction of digital development and to change the way we think when we think about news — a time for news media to transform from an outside observer into a change maker.

AI solutions for news media

If the news media wants to affect how news content is created, developed, presented and delivered to us in the future, they need to take an active role in AI development. If news organizations want to understand the way data and information are constantly affected and manipulated in digital environments, they need to start embracing the possibilities of machine learning.

But how can news media ever compete with today’s AI leaders?

News organisations have one thing that Google, Facebook and other big internet players don’t yet have: news organizations own the content creation process and thus have a deep and detailed content understanding. By focusing on appropriate AI solutions, they can combine the data related to the content creation and content consumption in a unique and powerful way.

News organizations need to use AI to augment you and me. And they need to augment journalists and the newsroom. What does this mean?

Augment the user-citizen

Personalization has been around for a while, but has it ever been designed and developed in the terms of news media itself? The goal for news media is to combine great content and personalized user experience to build a seamless and meaningful news experience that is in line with journalistic principles and values.

For news, the upcoming real-time machine learning methods, such as online learning, offer new possibilities to understand the user’s preferences in their real-life context. These technologies provide new tools to break news and tell stories directly on your lock screen.

An intelligent notification system sending personalized news notifications could be used to optimize content and content distribution on the fly by understanding the impact of news content in real time on the lock screens of people’s mobile devices. The system could personalize the way the content is presented, whether serving voice, video, photos, augmented reality material or visualizations, based on users�� preferences and context.

Significantly, machine learning can be utilized to create new forms of interaction between people, journalists and the newsroom. Automatically moderated commenting is just one example already in use today. Think if it would be possible to build interactions directly on the lock screen that let the journalists better understand the way content is consumed, simultaneously capturing in real time the emotions conveyed by the story.

By opening up the algorithms and data usage through data visualizations and in-depth articles, the news media could create a new, truly human-centered form of personalization that lets the user know how personalization is done and how it’s used to affect the news experience.

And let’s stop blaming algorithms when it comes to filter bubbles. Algorithms can be used to diversify your news experience. By understanding what you see, it’s also possible to understand what you haven’t seen before. By turning some of the personalization logic upside down, news organizations could create a machine learning-powered recommendation engine that amplifies diversity.

Augment the journalist

In the domain of abstracting and contextualizing new information and unpredictable (news) events, human intelligence is still invincible.

The deep content understanding of journalists can be used to teach an AI-powered news assistant system that would become better over time by learning directly from the journalists using it, simultaneously taking into account the data that flows from the content consumption.

A smart news assistant could point out what kinds of content are connected implicitly and explicitly, for example based on their topic, tone of voice or other meta-data such as author or location. Such an intelligent news assistant could help the journalist understand their content even better by showing which previous content is related to the now-trending topic or breaking news. The stories could be anchored into a meaningful context faster and more accurately.

Innovation and digitalization doesn’t change the culture of news media if it’s not brought into the very core of the news business.

AI solutions could be used to help journalists gather and understand data and information faster and more thoroughly. An intelligent news assistant can remind the journalist if there’s something important that should be covered next week or coming holiday season, for example by recognizing trends in social media or search queries or highlighting patterns in historic coverage. Simultaneously, AI solutions will become increasingly essential for fact-checking and in detecting content manipulation, e.g. recognizing faked images and videos.

An automated content production system can create and annotate content automatically or semi-automatically, for example by creating draft versions based on an audio interview, that are then finished by human journalists. Such a system could be developed further to create news compilations from different content pieces and formats (text, audio, video, image, visualization, AR experiences and external annotations) or to create hyper-personalized atomized news content such as personalized notifications.

The news assistant also could recommend which article should be published next using an editorial push notification, simultaneously suggesting the best time for sending the push notification to the end users. And as a reminder, even though Google’s Duplex is quite a feat, natural language processing (NLP) is far from solved. Human and machine intelligence can be brought together in the very core of the content production and language understanding process. Augmenting the linguistic superpowers of journalists with AI solutions would empower NLP research and development in new ways.

Augment the newsroom

Innovation and digitalization doesn’t change the culture of news media if it’s not brought into the very core of the news business concretely in the daily practices of the newsroom and business development, such as audience understanding.

One could start thinking of the news organization as a system and platform that provides different personalized mini-products to different people and segments of people. Newsrooms could get deeper into relevant niche topics by utilizing automated or semi-automated content production. And the more topics covered and the deeper the reporting, the better the newsroom can produce personalized mini-products, such as personalized notifications or content compilations, to different people and segments.

In a world where it’s increasingly hard to distinguish a real thing from fake, building trust through self-reflection and transparency becomes more important than ever. AI solutions can be used to create tools and practices that enable the news organization and newsroom to understand its own activities and their effects more precisely than ever. At the same time, the same tools can be used to build trust by opening the newsroom and its activities to a wider audience.

Concretely, AI solutions could detect and analyze possible hidden biases in the reporting and storytelling. For example, are some groups of people over-presented in certain topics or materials? What has been the tone of voice or the angle related to challenging multi-faceted topics or widely covered news? Are most of the photos depicting people with a certain ethnic background? Are there important topics or voices that are not presented in the reporting at all? AI solutions also can be used to analyze and understand what kind of content works now and what has worked before, thus giving context-specific insights to create better content in the future.

AI solutions would help reflect the reporting and storytelling and their effects more thoroughly, also giving new tools for decision-making, e.g. to determine what should be covered and why.

Also, such data and information could be visualized to make the impact of reporting and content creation more tangible and accessible for the whole newsroom. Thus, the entire editorial and journalistic decision-making process can become more open and transparent, affecting the principles of news organizations from the daily routines to the wider strategical thinking and management.

Tomorrow’s news organizations will be part human and part machine. This transformation, augmenting human intelligence with machines, will be crucial for the future of news media. To maintain their integrity and trustworthiness, news organizations themselves need to able to define how their AI solutions are built and used. And the only way to fully realize this is for the news organizations to start building their own AI solutions. The sooner, the better — for us all.

via TechCrunch

#IFTTT #TechCrunch

0 notes

doorrepcal33169 · 7 years ago

Text

Practical applications of reinforcement learning in industry

An overview of commercial and industrial applications of reinforcement learning.

The flurry of headlines surrounding AlphaGo Zero (the most recent version of DeepMind’s AI system for playing Go) means interest in reinforcement learning (RL) is bound to increase. Next to deep learning, RL is among the most followed topics in AI. For most companies, RL is something to investigate and evaluate but few organizations have identified use cases where RL may play a role. As we enter 2018, I want to briefly describe areas where RL has been applied.

RL is confusingly used to refer to a set of problems and a set of techniques, so let’s first settle on what RL will mean for the rest of this post. Generally speaking, the goal in RL is learning how to map observations and measurements to a set of actions while trying to maximize some long-term reward. This usually involves applications where an agent interacts with an environment while trying to learn optimal sequences of decisions. In fact, many of the initial applications of RL are in areas where automating sequential decision-making have long been sought. RL poses a different set of challenges from traditional online learning, in that you often have some combination of delayed feedback, sparse rewards, and (most importantly) the agents in question are often able to affect the environments with which they interact.

Deep learning as a machine learning technique is beginning to be used by companies on a variety of machine learning applications. RL hasn’t quite found its way into many companies, and my goal is to sketch out some of the areas where applications are appearing.

Figure 1. Slide courtesy of Ben Lorica.

Before I do so, let me start off by listing some of the challenges facing RL in the enterprise. As Andrew Ng noted in his keynote at our AI Conference in San Francisco, RL requires a lot of data, and as such, it has often been associated with domains where simulated data is available (gameplay, robotics). It also isn’t easy to take results from research papers and implement them in applications. Reproducing research results can be challenging even for RL researchers, let alone regular data scientists (see this recent paper and this OpenAI blog post). As machine learning gets deployed in mission-critical situations, reproducibility and the ability to estimate error become essential. So, at least for now, RL may not be ideal for mission-critical applications that require continuous control.

AI notwithstanding, there are already interesting applications and products that rely on RL. There are many settings involving personalization, or the automation of well-defined tasks, that would benefit from sequential decision-making that RL can help automate (or at least, where RL can augment a human expert). The key for companies is to start with simple uses cases that fit this profile rather than overly complicated problems that “require AI.” To make things more concrete, let me highlight some of the key application domains where RL is beginning to appear.

Robotics and industrial automation

Applications of RL in high-dimensional control problems, like robotics, have been the subject of research (in academia and industry), and startups are beginning to use RL to build products for industrial robotics.

Industrial automation is another promising area. It appears that RL technologies from DeepMind helped Google significantly reduce energy consumption (HVAC) in its own data centers. Startups have noticed there is a large market for automation solutions. Bonsai is one of several startups building tools to enable companies to use RL and other techniques for industrial applications. A common example is the use of AI for tuning machines and equipment where expert human operators are currently being used.

Figure 2. Slide from Mark Hammond, used with permission.

With industrial systems in mind, Bonsai recently listed the following criteria for when RL might be useful to consider:

You’re using simulations because your system or process is too complex (or too physically hazardous) for teaching machines through trial and error.

You’re dealing with large state spaces.

You’re seeking to augment human analysts and domain experts by optimizing operational efficiency and providing decision support.

Data science and machine learning

Machine learning libraries have gotten easier to use, but choosing a proper model or model architecture can still be challenging for data scientists. With deep learning becoming a technique used by data scientists and machine learning engineers, tools that can help people identify and tune neural network architectures are active areas of research. Several research groups have proposed using RL to make the process of designing neural network architectures more accessible (MetaQNN from MIT and Net2Net operations). AutoML from Google uses RL to produce state-of-the-art machine-generated neural network architectures for computer vision and language modeling.

Looking beyond tools that simplify the creation of machine learning models, there are some who think that RL will prove useful in assisting software engineers write computer programs.

Education and training

Online platforms are beginning to experiment with using machine learning to create personalized experiences. Several researchers are investigating the use of RL and other machine learning methods in tutoring systems and personalized learning. The use of RL can lead to training systems that provide custom instruction and materials tuned to the needs of individual students. A group of researchers is developing RL algorithms and statistical methods that require less data for use in future tutoring systems.

Health and medicine

The RL setup of an agent interacting with an environment receiving feedback based on actions taken shares similarities with the problem of learning treatment policies in the medical sciences. In fact, many RL applications in health care mostly pertain to finding optimal treatment policies. Recent papers cited applications of RL to usage of medical equipment, medication dosing, and two-stage clinical trials.

Text, speech, and dialog systems

Companies collect a lot of text, and good tools that can help unlock unstructured text will find users. Earlier this year, AI researchers at SalesForce used deep RL for abstractive text summarization (a technique for automatically generating summaries from text based on content “abstracted” from some original text document). This could be an area where RL-based tools gain new users, as many companies are in need of better text mining solutions.

RL is also being used to allow dialog systems (i.e., chatbots) to learn from user interactions and thus help them improve over time (many enterprise chatbots currently rely on decision trees). This is an active area of research and VC investments: see Semantic Machines and VocalIQ—acquired by Apple.

Media and advertising

Microsoft recently described an internal system called Decision Service that has since been made available on Azure. This paper describes applications of Decision Service to content recommendation and advertising. Decision Service more generally targets machine learning products that suffer from failure modes including “feedback loops and bias, distributed data collection, changes in the environment, and weak monitoring and debugging.”

Other applications of RL include cross-channel marketing optimization and real time bidding systems for online display advertising.

Finance

Having started my career as a lead quant in a hedge fund, it didn’t surprise me that few finance companies are willing to talk on record. Generally speaking, I came across quants and traders who were evaluating deep learning and RL but haven’t found sufficient reason to use the tools beyond small pilots. While potential applications in finance are described in research papers, few companies describe software in production.

One exception is a system used for trade execution at JPMorgan Chase. A Financial Times article described an RL-based system for optimal trade execution. The system (dubbed “LOXM”) is being used to execute trading orders at maximum speed and at the best possible price.

As with any new technique or technology, the key to using RL is to understand its strengths and weaknesses, and then find simple use cases on which to try it. Resist the hype around AI—rather, consider RL as a useful machine learning technique, albeit one that is best suited for a specific class of problems. We are just beginning to see RL in enterprise applications. Along with continued research into algorithms, many software tools (libraries, simulators, distributed computation frameworks like Ray, SaaS) are beginning to appear. But it’s fair to say that few of these tools come with examples aimed at users interested in industry applications. There are, however, already a few startups that are incorporating RL into their products. So, before you know it, you might soon be benefiting from developments in RL and related techniques.

Related resources:

“Reinforcement learning explained”

Deep reinforcement learning in the enterprise—Bridging the gap from games to industry (2017 Artificial Intelligence Conference presentation by Mark Hammond)

Ray: A distributed execution framework for reinforcement learning applications (2017 Artificial Intelligence Conference presentation by Ion Stoica)

Deep reinforcement learning for robotics (2016 Artificial Intelligence Conference presentation by Pieter Abbeel)

Cars that coordinate with people (2017 Artificial Intelligence Conference keynote by Anca Dragan)

“Neuroevolution: A different kind of deep learning”

“Why continuous learning is key to AI”

Continue reading Practical applications of reinforcement learning in industry.

from FEED 10 TECHNOLOGY http://ift.tt/2Apm7qG

#FEED 10 TECHNOLOGY

0 notes

repmywind02199 · 7 years ago

Text

Practical applications of reinforcement learning in industry

An overview of commercial and industrial applications of reinforcement learning.

Figure 1. Slide courtesy of Ben Lorica.

Robotics and industrial automation

Figure 2. Slide from Mark Hammond, used with permission.

With industrial systems in mind, Bonsai recently listed the following criteria for when RL might be useful to consider:

You’re using simulations because your system or process is too complex (or too physically hazardous) for teaching machines through trial and error.

You’re dealing with large state spaces.

You’re seeking to augment human analysts and domain experts by optimizing operational efficiency and providing decision support.

Data science and machine learning

Looking beyond tools that simplify the creation of machine learning models, there are some who think that RL will prove useful in assisting software engineers write computer programs.

Education and training

Health and medicine

Text, speech, and dialog systems

Media and advertising

Other applications of RL include cross-channel marketing optimization and real time bidding systems for online display advertising.

Finance

Related resources:

“Reinforcement learning explained”

Deep reinforcement learning in the enterprise—Bridging the gap from games to industry (2017 Artificial Intelligence Conference presentation by Mark Hammond)

Ray: A distributed execution framework for reinforcement learning applications (2017 Artificial Intelligence Conference presentation by Ion Stoica)

Deep reinforcement learning for robotics (2016 Artificial Intelligence Conference presentation by Pieter Abbeel)

Cars that coordinate with people (2017 Artificial Intelligence Conference keynote by Anca Dragan)

“Neuroevolution: A different kind of deep learning”

“Why continuous learning is key to AI”

Continue reading Practical applications of reinforcement learning in industry.

http://ift.tt/2Apm7qG

#FEED 10 TECHNOLOGY

0 notes

csemntwinl3x0a1 · 7 years ago

Text

Practical applications of reinforcement learning in industry

An overview of commercial and industrial applications of reinforcement learning.

Figure 1. Slide courtesy of Ben Lorica.

Robotics and industrial automation

Figure 2. Slide from Mark Hammond, used with permission.

With industrial systems in mind, Bonsai recently listed the following criteria for when RL might be useful to consider:

You’re using simulations because your system or process is too complex (or too physically hazardous) for teaching machines through trial and error.

You’re dealing with large state spaces.

You’re seeking to augment human analysts and domain experts by optimizing operational efficiency and providing decision support.

Data science and machine learning

Looking beyond tools that simplify the creation of machine learning models, there are some who think that RL will prove useful in assisting software engineers write computer programs.

Education and training

Health and medicine

Text, speech, and dialog systems

Media and advertising

Other applications of RL include cross-channel marketing optimization and real time bidding systems for online display advertising.

Finance

Related resources:

“Reinforcement learning explained”

Deep reinforcement learning in the enterprise—Bridging the gap from games to industry (2017 Artificial Intelligence Conference presentation by Mark Hammond)

Ray: A distributed execution framework for reinforcement learning applications (2017 Artificial Intelligence Conference presentation by Ion Stoica)

Deep reinforcement learning for robotics (2016 Artificial Intelligence Conference presentation by Pieter Abbeel)

Cars that coordinate with people (2017 Artificial Intelligence Conference keynote by Anca Dragan)

“Neuroevolution: A different kind of deep learning”

“Why continuous learning is key to AI”

Continue reading Practical applications of reinforcement learning in industry.

http://ift.tt/2Apm7qG

#FEED 10 TECHNOLOGY

0 notes