#ai interview copilot | Explore Tumblr posts and blogs

trylkstopocket · 4 years ago

Text

Papers with Code Newsletter #13

Welcome to the 13th issue of the Papers with Code newsletter. In this edition, we cover:

a summary of some of the latest ML papers in code generation and understanding,

a Vision-and-Language model with a CLIP-based visual encoder,

a deep learning algorithm (AlphaFold) for improving experimental accuracy of protein structures prediction,

an RL agent that combines contrastive learning and a hybrid LSTM-transformer architecture to improve data efficiency,

... and much more.

Trending Papers with Code 📄

Machine Learning for Code Generation and Understanding

Machine learning for code generation and understanding continued to make headlines these past few weeks. A recent example of this is the latest release of GitHub Copilot, an AI-based program synthesizer. Recently, there has been tremendous interest and progress in this area of research, including new methodologies, evaluation techniques, new benchmarks, large-scale datasets, and more.

In this issue, we feature some of the latest ML papers for code generation and understanding:

Evaluating Large Language Models Trained on Code

Example problem from the HumanEval dataset. Prompt provided to the model is shown with a white background and the model-generated completion is shown with a yellow background. Figure source: Chen et al. (2021)

Large language models have become general-purpose methods used for generation and representation in natural language processing, computer vision, audio and speech processing, and even in multi-modality problems. There is a huge interest to test and improve the program synthesis capabilities of popular language models. Chen et al. (2021) recently proposed a model called Codex, which involves a GPT language model fine-tuned on publicly available code from GitHub, and study its Python code-writing capabilities.

Main findings: The main task that Codex aims to solve is generating standalone Python functions from docstrings, and evaluating the correctness of the code samples automatically using unit tests. An evaluation framework with 164 original programming problems, HumanEval, is used to test whether the samples generated from the model pass some unit tests. Results show that with a single sample, a 12B parameter Codex can solve 28.8% of the problems. The programming problems assess language comprehension, algorithms, including some that are comparable to basic software interview questions. In contrast, a 6B GPT-J model achieves 11.4% while all other GPT models achieve near 0%. It's also interesting to note that a production version of Codex already powers GitHub Copilot.

Paper & Code

Improving Semantic Code Search through Multimodal Code Representation

Multimodal learning schematic diagram. Figure source: Gu et al. (2021)

Semantic code search is a task to retrieve the most semantically relevant code snippets for a given natural language query. Current state-of-the-art approaches focus on representing query and code data as vectors for semantic similarity computation. Gu et al. (2021) propose new methods, based on tree-serialized representations and multimodal code representations, to improve semantic code search.

What's new: This work uses tree-serialized representations to study whether they are more informative than token representation for neural code search. The method is designed to enrich the semantic representation of the program. The novel tree structure is shown to be an effective modality for multimodal code search. The authors show that traversal-based representations perform better than sampling-based representations. In addition, two quantification metrics are defined to separately measure the completeness of the syntactic and semantic information conveyed by various code representations. The proposed multimodal learning architecture is shown in the figure above.

Paper & Code

Recent ML Papers on Code Generation and Understanding

Below we feature some recent ML papers on the topic of code generation and understanding:

📄 A Dataset for Automatic Shellcode Generation - Liguori et al. (2021)

📄 Improving Extractive Code Generation - Orlanski and Gittens (2021)

📄 APPS - Hendrycks et al. (2021)

📄 Code-Text Transformer - Phan et al. (2021)

📄 CodeXGLUE - Lu et al. (2021)

📄 CodeTrans - Elnaggar et al. (2021)

📄 Programming Puzzles - Schuster et al. (2021)

📄 CodeBERT - Feng et al. (2020)

📄 PyMT5 - Clement et al. (2020)

📄 Unsupervised Translation of Programming Languages - Lachaux et al. (2020)

Find more of these works under our Code Generation task page.

AlphaFold...with Code!

Figure source: https://github.com/deepmind/alphafold

🎉

Open source code was released for DeepMind's AlphaFold yesterday!

Protein function can be inferred through their structure, which is why protein folding is an important task to solve. Two current approaches, physical interactions and history-based, have dominated protein prediction but they fall short of experimental accuracy which limits their utility. Jumper et al. (2021) recently proposed AlphaFold, a computational approach based on neural networks capable of protein structures prediction to near experimental accuracy.

What's New: AlphaFold leverages novel neural network architectures and training procedures. In summary, the proposed approach to improve accuracy of structure prediction involves: 1) an architecture that jointly embeds multiple sequence alignments (MSA) and pairwise features, 2) a new output representation and loss that improves end-to-end structure prediction, an equivariant attention architecture, a mechanism to iteratively refine predictions using intermediate losses, masked MSA loss to jointly train with structure, self-distillation to learn from unlabelled protein sequences, and self-estimates of accuracy. Overall, AlphaFold incorporates physical and biological knowledge about protein structure and achieves competitive results in the challenging CASP14.

Paper & Code

Can CLIP Benefit Vision-and-Language Tasks?

The typical training process of a V&L model. Figure source: Shen et al. (2021)

Vision-and-Language (V&L) tasks such as visual question answering (VQA) aim at understanding the semantics of the visual world with the help of natural language. The visual encoder in these systems has been found to be a performance bottleneck. One of the limitations stem from a dependency on manually-annotated datasets where class labels are pre-defined thus limiting visual representation. The visual encoder can benefit from more diverse and large-scale data sources. A good example is the recent CLIP model that learns visual concepts with language supervision. This model is trained on 400M noisy image-text pairs crawled from the internet. CLIP benefits from this data collection process which could also bring similar benefits to V&L tasks. To address this, Shen et al. (2021) recently proposed to integrate CLIP with existing V&L models by replacing the visual encoder with CLIP's visual encoder.

Main idea: Recent studies have shown that directly applying CLIP to V&L tasks is difficult because these tasks require complex multi-modal reasoning. Two scenarios are considered: 1) plugging CLIP into direct task-specific fine-tuning; 2) integrating CLIP with V&L pre-training on image-text pairs and transfer to downstream tasks. In summary, the proposed method (CLIP-ViL) uses CLIP as the visual encoder for the V&L model. Results show that CLIP significantly outperforms widely-used visual encoders that rely on in-domain annotated data. CLIP-ViL achieves competitive results on V&L tasks such as visual question answering and visual entailment.

Paper & Code

CoBERL

An overview of the CoBERL architecture. Figure source: Banino et al. (2021)

In a previous issue of the newsletter, we featured the Decision Transformer which aimed at showing that the reinforcement learning problem could be casted as conditional sequence modeling. Recently, more investigations have been conducted to bring the benefits of sequence modeling to the realm of reinforcement learning. For instance, Banino et al. (2021) recently presented a model, CoBERL, that aims to improve an agent's performance across a wide range of environments and tasks by leveraging architectural improvements and a novel contrastive representation learning objective.

What's new: The first contribution of this work is a novel contrastive representation learning objective that improves representations by enforcing self-attention consistency in the prediction of masked inputs. Secondly, CoBERL combines the strengths of two forms of memory representations: the all-to-all attention of transformers and sequential processing of LSTMs. Overall, the combination of the hybrid LSTM-transformer architecture and new contrastive loss aims to improve data efficiency, and avoid the need for hand engineered data augmentations. Results show that CoBERL improves performance across the full Atari suite.

Paper & Code

Trending Libraries and Datasets 🛠

Trending datasets

UrbanScene3D - a large-scale urban scene dataset associated with a simulator, which consists of both man-made and real-world reconstruction scenes in different scales.

iMiGUE - a dataset for emotional artificial intelligence research, which focuses on nonverbal body gestures without using any identity information.

TrajAir - a dataset containing aircraft trajectories in an untowered terminal airspace collected over 8 months.

Trending libraries/tools

DRIFT - a toolkit that allows researchers to track research trends and developments over the years.

LightSeq - a highly efficient inference library for models in the Transformer family.

Mava - a research framework for distributed multi-agent reinforcement learning.

Community Highlights ✍️

We would like to thank:

The OGB team for collaborating on adding the leaderboards for the Open Graph Benchmark (OGB) to Papers with Code.

@agoyal for contributing to Datasets and Leaderboards, including the addition of the Rel3D dataset for grounding spatial relations in 3D.

@rogerdjq for several contributions, including many updates to the Aspect Term Extraction and Sentiment Classification task.

@Abrar2652 for contributing to Dataset, including the addition of the Narvik Road Dataset.

@enk100 for contributing to several benchmarks, including many additions to the Speech Separation on WSJ0-4mix leaderboard.

@sailorzhang for many contributions, including updates to the Audio-Visual Active Speaker Detection on AVA-ActiveSpeaker leaderboard.

Special thanks to the hundreds of other of contributors for their ongoing contributions to Papers with Code.

---

We would be happy to hear your thoughts and suggestions on the newsletter. Please reply to [email protected].

See previous issues

Join us on Slack and Twitter

You are receiving this e-mail because you subscribed to the weekly digest on paperswithcode.com. Click here to unsubscribe

Follow on twitter: @paperswithcode

0 notes

jenessa-mercier · 8 years ago

Text

First Session [Pacific Rim AU Drabble]

PAN PACIFIC DEFENSE CORPS: PERSONNEL DOSSIER

NAME:Jenessa Mercier

ASSIGNED TEAM: Rangers; ID J-SLM 117.12-D

DATE OF ACTIVE SERVICE: December 17th, 2021.

CURRENT SERVICE STATUS: Standby pending Pilot certification and procurement of co-pilot.

BIOGRAPHY: Born on October 16th, 1992. Eldest daughter of Javed Mercier, chief engineer of the second generation of the Jaeger program.

COMMAND ASSESSMENT: Trained in a variety of martial arts, Mercier has often employed them with her past time in bar fights before her training. Cocky and brash at times but has the skills and adaptability to match. Graduated the academy with honors, her skills include long range marksmanship and knife work. Outside life is unremarkable save for her penchant for engaging in underground fights to push her skills further. Her blunt attitude causes friction between the newer recruits but the more experienced know her insight comes from experience.

NOTES: Has a record of 13 sim drops, 12 kills. Co-pilot drifting compatibility has been problematic. Only one successful Drift to date but former partner was teamed up with a newer recruit and assigned a Mark IV, Blooming Field. Mercier has been on standby ever since.

Her eyes were already closed as she stood in the cockpit of the Jaeger they would be using as a test for their link. While an old beast by generation standards, the Mark III Mistral Vanguard was still a powerful weapon in the right hands and while the nuclear aspect was frowned upon now, the dependability that went into the older series was nothing to scoff at. The ranger took a deep breath as the sound of the technicians approaching drew her out of her thoughts, the woman slipping off her jacket as the suits were produced. It took a few minutes as Jenessa was outfitted with her Pilot suit, the black and silver armor being carefully bolted on. She was well accustomed to it by this point, her head tilting slightly as the back pieces were secured just a bit too snugly. Accepting the helmet offered to her, she was left little to do but to wait on the right side for her copilot to board whilst the technicians ran their usual checks. Thankfully it was a short wait as Ranger Alicimos was precisely on time, already outfitted and looking to be the model pilot the world seemed to be in love with. It was the sort of look that would get interviews after a promising start, maybe even a sponsorship if they were especially lucky. It was the usual type of thing that Jenessa disliked even if it was necessary, wasting time on public appearances when there was a war to be fought. “You’re here early.” Aven noted, taking his spot on the left side after a moment’s hesitation. His helmet was tucked under his left arm, the armor gleaming in the lights of the hangar filtering in. The hangar itself was abuzz with activity, those stationed there were always curious to see a new Pilot pair come together and be granted a Jaeger. Even know, there were groups of nonessential personnel gathering on the catwalks and railings to watch. “Do you have any experience piloting?” Aven asked, drawing a glance from Jenessa as they were motioned to step into place. “If we’re going to be working together, it’d be good to know. So I know I’m not paired up with someone who isn’t completely new at this.” “I’ve been piloting for over a year, rookie. You’re fresh out of the academy and into the Pilot program. Let me offer you a piece of advice...” Jenessa slipped her helmet on and felt the distinct thrum of machinery come to life beneath her. “Don’t ask questions that you’ll get the answers to once we’re Drifting.” He shot her an annoyed look, copying her and placing his helmet on. The clamps beneath them locked onto their legs, securing them in place as their helmet’s speakers came to life. The remainder of the technicians finished their check and after a nod to the pair, exited the cockpit. “Alrighty boys and girls this is Tendo speaking. We have your regularly scheduled Neural Sync test for today and I do hope you’ve secured all your belongings properly for this. The Jaeger Program is not responsible for any lost or stolen-” “Tendo.” The stern voice of Pentecost cut through their operator’s voice, the sound of Tendo clearing his voice before adopting a professional tone. A faint smile played across Jenessa’s lips as they went through their briefing on their testing for today. A simple sync test and a neural handshake, nothing more. If they were able to achieve a high sync rate, proper designation and deployment would follow. There was the sudden rush of vertigo as they were dropped at a quick pace, the pair buckling instinctively at the sensation of a pseudo drop. The sudden stop and sound of rotating metal, various sounds of connecting machinery as Tendo’s voice came through again. “Preparing for neural handshake. Don’t chase the rabbit, I’m sure you’re aware. Initiating in 3.... 2....” Jenessa closed her eyes as Aven did the same beside her, their heads bowed. “1.” The sensation wasn’t wholly unfamiliar to Jenessa but the rush of memories that flooded her mind’s eye were. A simple childhood of a young boy playing with dinosaur toys... Visiting a mother’s grave... Pride at his father defending humanity in a Mark I Jaeger... Grief when he died of cancer but still there was that pride... Jenessa pulled herself out of the memories with a deep breath, opening her eyes as the sensation lessened to a dull tickle in the back of her head. Taking a moment to spare a glance at Aven, the rookie had a hand up to his helmet as their comms line came back to life. “Neural handshake is holding but we’re getting fluctuations on Alcimos’ end. What seems to be the issue, Ranger?” Pentecos’s voice seemed to draw Aven out of his slump, the pilot looking lost as he looked around the cockpit. “Give him a minute, sir. Rookie pilot and he doesn’t have a history of drifting. I’m sure it’ll stabilize in a moment.” Jenessa kept a close eye on him as his hand went to his helmet again, the female Pilot’s eyes narrowing somewhat. “Sir, neural handshake is dropping rapidly. Alcimos is losing-” Suddenly the entire Jaeger shifted and there was a sudden twinge of unease in Jenessa’s shoulder as they were thrown out of alignment. The AI voice of Mistral came online, informing them of a Pilot disconnect. Shuddering in pain as the entire Jaeger cortex was shifted onto her, Jenessa felt her knees buckle again before her comms exploded. “Alcimos, put your helmet back on! We are in the middle of a neural handshake and- Tendo, emergency shut off! I don’t need to lose anyone today.” Jenessa was aware of the voices but the utter strain on her body from the Jaeger interface was oppressing for several long moment before it suddenly lifted, the woman gasping in relief at the loss. Gasping as tech crews scrambled towards the now inert Jaeger, Jenessa spared a glance at Aven who had dropped his helmet and was holding his head in his hands. While she longed to snap at him, the more seasoned ranger sighed and carefully removed her own helmet. It wouldn’t be the first time this had happened. The intercom of Mistral herself came to life, running on auxiliary power as Commander Pentecost came in. “Both Rangers Mercier and Alcimos report to me within the hour for a debrief on what just happened.”

#Pacific Rim AU #drabble

8 notes · View notes

spycamp · 7 years ago

Text

From: Alex Erickson Sent: Saturday, May 26, 2018 9:20 AM To: [email protected] Subject: Advanced Copy

I referenace A>I>M>E> on this and with ai project time can be of the essnce. I am very pleased with how I have framed my future relationship with NASA in this request for a military briefing.

DHSHDS a Military Briefing Folder Request

Alexander William Erickson Born June 29

AWE629 an American Spy

Family History

In the year 2012, I was married then, I travelled to Savannah, GA to be spend time with the in-laws. The week was spent with family with conversations about my career plans and application to Department of Homeland Security (D.H.S.) were discussed regularly. My wife and me could not get along during the summer of 2012, and she travelled back to Savannah to be with family without me. During the summer I cancelled the interview process with D.H.S. and planned my career into real estate sales. While this career transition ringed in the ears of people I talked too, the plans felt flat, and I felt like I was going through a process of failure not preparing for success in a new career. I had lied to my parents about applying for government police work, and I needed to mature in order to understand myself.

In October 2013, I had been divorced for almost a year, but I had become certain while travelling in Virginia that I was in fact an undercover spy unwinding from self-awareness. I immediately began self-improvement by deciding to apply to Stanford University for a doctorate in Communications and decided that if that failed to work I would fix The Santa Clara County Mental Health System. I was under physical duress from clandestine communications and I needed to order the voices in my head.

I have spent these last five years preparing a plan as New Leadership and will continue to approach The Department of Homeland Security as part of my family.

I require an update about my family of contacts at DHS included in the DHSHDS folder briefing.

The Port Authority

My brother-in-law works for DHS-ICE and when I visited Savannah and he spent the day with the marine port authority handling possible illegals trying to land in the Georgia Port. In my assessment I was Recruited as a spy by my brother-in-law when I visited him in Georgia. My work as a spy has come to two conclusions. First, a new Port of Entry with a weapons and communications advanced manufacturing facility is necessary for security purposes as we pass through into an advanced era of technology. Second, there is a tunnel designed for artificial intelligence and Federal Farming that is being used presently as a gangland homeland slave tunnel. I am planning the new port of entry for The United Nations with a top-secret security briefing, and this intra-state tunnel of rumors will be secured for a major artificial intelligence project called AIME: Artificial Intelligence Mapped Endeavors.

The sociology of this project is to develop a family atmosphere of support for me at The Port Authority and D.H.S. overall. As a corporation of people D.H.S. will be involved in a “relationship fox hole” with Me and Angelina Jolie. Angelina Jolie is in a fox hole with The United Nations. I am going to be spending one year travelling the world with Angelina Jolie, and a lot of the time will be spent with The United Nations. I look forward to getting to spend time with D.H.S. to plan A.I.M.E. and GLYNCO2 while travelling.

The trip is called The Ranger Wedding Period. I am going to be completing an economic slavery and spy survey. Other possible participants include Cytherea, Cytherea and Alexander will be training to operate a Global Live Security Agency, or in other words a social embroidery. Alexis Texas Thea Samper will be travelling with NASA support, we will be training as copilots for NASA’s componentry program and my IBM Rocket Ship program.

The Sociology and Cooperation of The Port Authority concerning me and my operations, any Requests for Information included in this section of the folder.

The 89 on Logic and Reasoning

I scored an 89 on the Logic and Reasoning test for The Department of Homeland Security. Ever since then, often times, I have found myself acting submissively against the greater will of logic and reason at times. And my understanding of that phenomena is that these plans for my career are very important, so important that my Handlers at D.H.S. expect me to lose everything except this quest to become The International Head of The CIA with a career as New Leadership.

Tests, Government Forms, Interviews, Details, please add into this section.

The Port of Entry Overview

This era is a transition from the modern era on Planet Earth to an advanced era of technological advancements in biology and manufacturing that require an elite method to create and provide security for personnel that regularly travel risky trade routes. Enough technology exists that a bag with communications gear and weaponry can be swapped out for scanned gear with biologically infused metal that looks the exact same but creates biological interferences with use. This method of terrorism can spread psychosis like a biological virus deconstructing a normal person’s psychology. With that understood I have began developing a Port of Entry as GLYNCO2 a completely new look at weapons and communications security.

The Port of Entry has a specific series of uses: First, the facility manufactures the Department of Homeland Security Utility Belt and tool kit. The Port of Entry will manufacture new radios, cell phones, handcuffs, leathers, keys, and handguns only for The Department of Homeland Security. This process which includes providing DNA as an input into the manufacturing process will provide every DHS Employee a Permanent Credit / ATM Card in addition to their new tools and utility belt.

This tool kit is under development to provide assurances that a Federal Police Officer on patrol for A.I.M.E. will not only secure the above ground perimeter but make the necessary underground adjustments while providing stability and peace of mind to the officer. Developing a mobile data communications system that is able to secure an intra-state underground advanced era farming systems for The Federal Government. Most importantly this new utility belt system will provide the best communications and family security for personnel available.

The Department of Homeland Security will complete its first complete round of manufacturing and distribution for The D.H.S. Police Personnel. Then D.H.S. will invite The United Nations into the new United Nations Port of Entry into The United State of America.

Please use this section of the brief to address questions and concerns about GLYNCO2.

Relationship with DHS & United Nations – The Germination of the Idea

I remember beginning 2018 by reading an article about Angelina Jolie meeting with a solider requesting military aide. The entire construct of the event seemed to have gone sour from the article. A meeting that obviously some of the handlers of Angelina Jolie and some the staff of the military officer never intended for this meeting to be fruitful. That caused me to look at my entire operation with The Department of Homeland Security having similar problems. My handlers at DHS, my former brother-in-law never expects me to perform and deliver a clandestine spy.

This operation may be about providing DHS as civilian leadership cover for me to build a relationship with The United Nations and The United States Army. Please address any developmental ideas about improving relationships between The US ARMY, The UN, and DHS in this section.

The United Nations Security Brief

I am planning a three week series of movements prior to the Ranger Wedding Period. Teasers Week is a road trip for collecting everyone for the Ranger Wedding Period. Boston City Planning will be the public relations and human resources week. New York Shopping week will include two fashion spreads, Alexander and Angelina went shopping with Vanity Fair, and the Alexis Texas VOGUE NASA interview. Friday of the New York Shopping Weeks I will speak at The United Nations, and that Saturday there is a meeting at The New York Stock Exchange for delivery of precious stones from my Trust Fund, The Lyons Long Term Trust. The United Nations Security Brief will get delivered during Boston City Planning Week.

I am developing a Global Live Security Agency. The first test of the security system will be to distribute this Security Brief. I am going to be using cyclists from New York to travel the world by plane, train, bus, and bicycle to deliver these security briefs to every principal member of The United Nations. I a recruiting the Federal Bureau of Investigation for supervision of this distribution.

The United Nations Security Brief will provide ambassadors, employees, and member citizens a complete summary of opportunity for the New Port of Entry. The Security Brief will address that the first pass through the Port of Entry will occur in military formation and will confirm an updated and finalized roster of members of The United Nations as recognized by The Department of Homeland Security. This is a security briefing not a security event. The briefing will tell everything we can disclose. The date for the event and travel details will be decided when The United Nations has sent a troupe to meet with the top-secret workers on the project.

The Department of Homeland Security will be preparing this top-secret briefing and I will read it with Angelina Jolie in Boston for the first time while we wait for responses from others who review the security briefing.

Please use this section to ask me questions about the briefings contents and distribution plans.

The United Nations Pager and Credit Card

I am not prepared to explain the reasons for the security measures. Someone at DHS GLYNCO2 will prick fingers collecting blood and will supply you a pager and credit card inked with United Nations Artificial Intelligence Forensics Tracking and your own blood. The pager holds the credit card.

I need to be briefed on oneway or twoway options for the pager.

The Ranger Wedding Period To Do List

I want to take this time out and get personal. I am your superhero. I am The DHS Superhero. And this is about improving socializing between UN & DHS. The United Nations have a superhero her name is Angelina Jolie. We are creating a very effective power couple with me & her. I am going to travel the world creating a new dictionary for Congress with her and the UN. I am The DHS Superhero.

Department of Homeland Security please call NASA and The Department of Defense and request to be briefed on The Nu Elemental Worker Unit Currency. And brief them about superhero socialization. Then prepare a brief for me on The Nu Elemental Worker Unit Currency from the superhero Alexander & Angelina perspective for me to discuss with her during a legal custody conference during Boston City Planning Week. I may need to offer her private education for her children while travelling, and this is how I afford such things.

My personal staff does not cover my artificial intelligence project during the Ranger Wedding Period. Any movement or effort for A.I.M.E. during the Ranger Wedding Period needs to be developed by DHS with Central Atlanta Progress from Atlanta, Georgia. I am planning a lot of Federal Employment around A.I.M.E. and a lot of personal business opportunities around A.I.M.E. in Atlanta and at ATL. This is an edgy operation and will include a Successful Drinkers of The Court system that will provide a social outlet for people in recovery that want to keep a drink or two on their personal calendar during recovery in partnership with a rental car company and ATL. I am very passionate about my Atlanta Hustle projects.

#cameragirlcollege

I am founding an elite college in Marseilles, France and its gathering identity is #cameragirlcollege. I am doing so in haste to be in association with this new Port of Entry for The United Nations. Members of my college will be the last in line to Roster through this Port of Entry secured by The Department of Homeland Security and will receive a device that does sms calendaring and credit card payments.

The college specialty is advanced engineering and biology projects creating diplomas from studies at multiple Universities, Federal Facilities, and Corporations. The college is interested in recruiting educated soldiers for advanced degrees at individual Universities.

A.I.M.E. The Atlanta Hustle of Alexander

The Federal Government is required to maintain reserves and harvests of popular and necessary goods. The problem from the outset of artificial intelligence is the liability of collecting human anatomy for production inputs. The Federal Government is going to regulate the artificial intelligence business by developing A>I>M>E> a road map for innovation that is a vertically integrated a technically sound singular structure of spy intelligence that is the only solution that wins for NASA beyond NASA. Which is winning the intergalactic space race. My plans in New Jersey with #AUTOCADCAPACITY @GAMECARTCORP for INTERTECHT LLC create the planetary balance to eventually win the intergalactic space race for NASA at NASA with A>I>M>E> equipment, my equipment.

The briefing on A.I.M.E. is that my Atlanta Hustle uses a slave dialect method for producing advanced results. I want this space open for ideas on how we approach that aspect of the business. From what I understand about the A.I.M.E. equipment is that it was used as a slave tunnel and therefore functions within a slave dialect. The public relations framework on this project needs to be thought through.

I like the backwoods culture, and if the culture is the problem, we will shut it down after we win in about thirty years and rebuild the entire thing.

AUTOBOUNDESCROW & AUTOCADESCROW

Prior to completing the surveys of The Ranger Wedding Period and #AUTOCADCAPACITY I want to set up AUTOBOUNDESCROW.

My explanation for autoboundescrow versus autocadescrow: With my NASA mission, Mission #BlackbirdSalad I am going to roll a SR71 Blackbird into my top-secret manufacturing facility and autoboundescrow will capture data confirming the entire planes contents and densities and placements. And NASA-CIA will review the contents and confirm the formation of an autoboundescrow account. I will then input my autocadescrow account information into the top-secret manufacturing system and days later I will roll out the plane for #BlackbirdSalad NASA Mission. We will photograph the plane, video and audio capture the plane, we will test fly the plane, and we will complete the mission to Mars. The vehicle will land, and we will roll it back into the top-secret manufacturing facility and the autocadescrow file will input the necessary changes to meet with the specifications of the autoboundescrow. And days later we will roll out the same SR71 Blackbird that was rolled in.

Autocadescrow and Autoboundescrow are going to be GROUNDFLOOR corporations from Atlanta, GA and may be destined to be words in the dictionary rather than company names.

Please brief me on the progress and plans updates.

________________________

Alexander William Erickson

#AWE629 #SCOTUS #WIPO #BankofJapan #America #AtlantaHusle #CEP #CentralAtlantaProgress

0 notes

adamgdooley · 7 years ago

Text

Melissa Di Donato, SAP: On intelligent ERP and how AI can change society

Melissa Di Donato

When Melissa Di Donato, chief revenue officer at SAP S/4HANA Cloud, was going through the interview process at Walldorf, the company’s home town, she was asked whether she had been there before. Di Donato took the opportunity to reveal the SAP watch she was wearing, issued the better part of two decades ago, which showed she was a certified consultant through her work at PwC.

While the company had never appeared on her CV per se, starting with a dean at business school who suggested getting into ‘this SAP thing’ as it ‘may catch on’, through consultancy, to a hosting partner, to even the other side of the fence with IBM, Oracle – “as one does” – and Salesforce, SAP has been very close throughout Di Donato’s career.

Hence moving to the company towards the end of last year, to help accelerate the cloud ERP business, felt like coming home.

The message is now around ‘intelligent cloud ERP’. As this publication reported in February, SAP is looking to S/4HANA, its public cloud offering, and adding artificial intelligence (AI) to ERP, helping customers to adjust and adopt business processes based on real-time data and insight.

Earlier this month, this went up a notch with the general release of CoPilot, a ‘digital assistant for the enterprise’ as the company’s annual SAPPHIRE conference. Bernd Leukert, executive board member, summed up SAP’s roadmap for attendees. “Openness is the game-changer, driving new business value,” he said. “Openness in every part of the company allows you to reach completely new levels of business process excellence.

“You can inject data and derive insights across every layer of your company, like traffic and other data into logistics, machine learning data into your maintenance plans, and social data from your customer profiles.”

Di Donato puts it this way. “It’s a big, big release for us, and a change I think to the industry,” she tells CloudTech. “Effectively, [CoPilot] learns to become aware of business context, is able to drive efficient collaboration, quickly is able to recognise and connect to business objects, and offers in-context chatting. It’s fast learning and embeds a lot of machine learning, and a lot of new innovative technology that’s intelligent… [to] truly transform the way ERP is.”

SAP illustrates the product’s capabilities in various videos. One looks atmonitoring and extending purchase contracts. Di Donato says that by 2020, the company is aiming for many financial functions to become ‘completely automated’. “At the end of the day, when I go out and speak with customers and say to them ‘what are you looking for in ERP?’, what can we give you to transform your business besides being less expensive, being less expensive is not the crux of where our customers are looking for benefit,” she says.

“The international finance reporting standards requirements are so extensive and so time-consuming that something basic – i.e. I want to close my books – is quite complex when you’re dealing with international reporting standards,” Di Donato adds. “Being able to offer faster time to close your books, taking away a lot of the functions and steps necessary with machine learning is hugely beneficial…detailed data in context to be able to make fast decisions…those are real business benefits that folks are seeing and asking for.”

Mention artificial intelligence in a business context, and while much of the conversation will discuss the potential benefits and efficiencies, another thought naturally occurs: will it affect my job? Back in January, Japanese company Fukoku Mutual Life Insurance made headlines after announcing it would replace more than 30 of its staff with IBM’s Watson Explorer AI.

The more optimistic view, as cited by respondents to an Adecco survey in April, is that AI will remove the duller, less palatable tasks and free up employees to be more creative. It’s a view with which Di Donato agrees. “It will be beneficial, and do I think they’ll be letting people go and that there’ll be a huge turn of events, saving money by not having people in the company? No, that’s not how we think necessarily at all,” she says.

“What we want to do is be able to allocate the mundane tasks to machines and machine learning technology, whilst taking the people and making them much more impactful on the business…giving them strategic roles, high impact roles, roles that can touch customers and service customers in a way that a machine can’t. It’s a reallocation of intelligence, if you will.”

Moreover, the rise of AI may represent a great opportunity for enhancement. Di Donato frequently speaks with young people, and the consistent message of the earlier the better in learning technical skills comes through. “I think we have the ability as a society to increase the value that everyone has to bring and bear,” she says. “When we take the mundane tasks away, it’s a perfect opportunity for our society to be upskilled and to learn more.

“Whilst yes, 99% of jobs are ‘mundane’, I venture to believe that not 99% of people want to do mundane tasks. We just have to do them, and we get used to them.”

SAP is also looking at other emerging technologies, with many coming under the banner of Leonardo, an innovation portfolio. This includes blockchain and the Internet of Things (IoT) – or not quite, as Di Donato explains. “I came on board and I was like ‘yawn’, I’m so sick of hearing about IoT,” she explains. “It’s not the Internet of Things anymore anyhow, it’s the Internet of Everything.

“What is true is that there are components of the Internet of Everything, IoT, machine learning, that the enterprise has not been able to consume,” she adds. “The problem was everyone talked a good game, but it wasn’t anywhere being used, and I think the difference now is that it was educational for 10 years, and now it’s consumption.

“We went from hypothesising about the business benefit of what IoT, and machine learning, and AI, and all of these things can do for the enterprise, and then all of a sudden it’s become embedded into our ERP.”