/* ---- Google Analytics Code Below */

Thursday, July 23, 2020

Alexa-Live Sessions

Attended this, mostly either broadly general or deep technical.  Dev and Device oriented.  Like the direction, lots to do  The on-demand pieces are at the link below, go to the bottom of the page at the link to register and see/hear the sessions:

Discover the Latest in Voice Technology
During Alexa Live, our virtual developer event, we brought together the developer and device maker community to explore the latest advancements in voice technology and Alexa development tools. Whether you build Alexa skills, make Alexa devices, or lead a business that’s incorporating voice, Alexa Live offered content and resources to help you build delightful customer experiences. View the on-demand sessions and resources to catch up on the news and technical deep-dives we shared at Alexa Live 2020. .... ' 

Lego Data Talk

Lego is a favorite company for a number of reasons, met with their management in the 90s. Here an upcoming talk via Retailwire, plan to participate.

LEGO: Brick by Brick, Built on the Foundation of Data
 Expert Webinar on July 30th, 2020:  11am EST | 8am PST 

More info and Register Now

Discover one of LEGO's fundamental bricks to commercial success: high-quality data.

In this webinar, hear from LEGO's Head of Experience, Torben Nielsen, on how the innovative brand worked with Loqate to enhance their user experiences, optimize business efficiencies, and improve the accuracy of 25% of their customer database.  ... "

Image GPT

Been looking at GPT from OpenAI, and at their site found Image GPT.   There a number of links to technical papers at the pieces below:

OpenAI first described GPT-3 in a research paper published in May. But last week it began drip-feeding the software to selected people who requested access to a private beta. For now, OpenAI wants outside developers to help it explore what GPT-3 can do, but it plans to turn the tool into a commercial product later this year, offering businesses a paid-for subscription to the AI via the cloud. ... 

Image GPT
We find that, just as a large transformer model trained on language can generate coherent text, the same exact model trained on pixel sequences can generate coherent image completions and samples. By establishing a correlation between sample quality and image classification accuracy, we show that our best generative model also contains features competitive with top convolutional nets in the unsupervised setting.

Introduction
Unsupervised and self-supervised learning,1 or learning without human-labeled data, is a longstanding challenge of machine learning. Recently, it has seen incredible success in language, as transformer2 models like BERT,3 GPT-2,4 RoBERTa,5 T5,6 and other variants78910 have achieved top performance on a wide array of language tasks. However, the same broad class of models has not been successful in producing strong features for image classification.11 Our work aims to understand and bridge this gap.

Transformer models like BERT and GPT-2 are domain agnostic, meaning that they can be directly applied to 1-D sequences of any form. When we train GPT-2 on images unrolled into long sequences of pixels, which we call iGPT, we find that the model appears to understand 2-D image characteristics such as object appearance and category. This is evidenced by the diverse range of coherent image samples it generates, even without the guidance of human provided labels. As further proof, features from the model achieve state-of-the-art performance on a number of classification datasets and near state-of-the-art unsupervised accuracy[1]  ...  " 

Responsible Innovation

Google reports on what they believe is responsible.   Considerable piece.   Lots of words.

An update on our work on AI and responsible innovation
Kent Walker  SVP, Global Affairs
Jeff Dean   Google Senior Fellow and SVP, Google Research and Health
Published Jul 9, 2020

AI is a powerful tool that will have a significant impact on society for many years to come, from improving sustainability around the globe to advancing the accuracy of disease screenings. As a leader in AI, we’ve always prioritized the importance of understanding its societal implications and developing it in a way that gets it right for everyone. 

That’s why we first published our AI Principles two years ago and why we continue to provide regular updates on our work. As our CEO Sundar Pichai said in January, developing AI responsibly and with social benefit in mind can help avoid significant challenges and increase the potential to improve billions of lives. 

The world has changed a lot since January, and in many ways our Principles have become even more important to the work of our researchers and product teams. As we develop AI we are committed to testing safety, measuring social benefits, and building strong privacy protections into products. Our Principles give us a clear framework for the kinds of AI applications we will not design or deploy, like those that violate human rights or enable surveillance that violates international norms. For example, we were the first major company to have decided, several years ago, not to make general-purpose facial recognition commercially available. ... " 

Meshing and Simulation

Design meshes are  a means for determining the design of simulation, especially for design problems.  Good discussion.

Better simulation meshes well for design software (and more)
New work on 2D and 3D meshing aims to address challenges with some of today’s state-of-the-art methods.   By Adam Conner-Simons | MIT CSAIL

The digital age has spurred the rise of entire industries aimed at simulating our world and the objects in it. Simulation is what helps movies have realistic effects, automakers test cars virtually, and scientists analyze geophysical data.

To simulate physical systems in 3D, researchers often program computers to divide objects into sets of smaller elements, a procedure known as “meshing.” Most meshing approaches tile 2D objects with patterns of triangles or quadrilaterals (quads), and tile 3D objects with patterns of triangular pyramids (tetrahedra) or bent cubes (hexahedra, or “hexes”).

While much progress has been made in the fields of computational geometry and geometry processing, scientists surprisingly still don’t fully understand the math of stacking together cubes when they are allowed to bend or stretch a bit. Many questions remain about the patterns that can be formed by gluing cube-shaped elements together, which relates to an area of math called topology.

New work out of MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) aims to explore several of these questions. Researchers have published a series of papers that address shortcomings of existing meshing tools by seeking out mathematical structure in the problem. In collaboration with scientists at the University of Bern and the University of Texas at Austin, their work shows how areas of math like algebraic geometry, topology, and differential geometry could improve physical simulations used in computer-aided design (CAD), architecture, gaming, and other sectors.

“Simulation tools that are being deployed ‘in the wild’ don’t always fail gracefully,” says MIT Associate Professor Justin Solomon, senior author on the three new meshing-related papers. “If one thing is wrong with the mesh, the simulation might not agree with real-world physics, and you might have to throw the whole thing out.”  .... ' 

In one paper,  https://diglib.eg.org/handle/10.1111/c 4074a gf14074a team led by MIT undergraduate Zoë Marschner developed an algorithm to repair issues that can often trip up existing approaches for hex meshing, specifically.  ..." 

Wednesday, July 22, 2020

How Good is Language Generator GPT-3?

Trying to understand the full range and usefulness of this, will schedule a test.   I did take a look at GPT-2 for a client.   Much more at the link ...

OpenAI’s new language generator GPT-3 is shockingly good—and completely mindless
The AI is the largest language model ever created and can generate amazing human-like text on demand but won't bring us closer to true intelligence.

by Will Douglas Heaven in TechologyReview
July 20, 2020

“Playing with GPT-3 feels like seeing the future,” Arram Sabeti, a San Francisco–based developer and artist, tweeted last week. That pretty much sums up the response on social media in the last few days to OpenAI’s latest language-generating AI.    ... "

Linkedin Plans to Lay off

Bell weather activity,especially when you would expect them to thrive in an era of increased job hunting and HR activity.

In A Scary Sign Of The Times, LinkedIn Plans To Lay Off 960 People    By Jack Kelly in Forbes

 In a sign of the times, with a little bit of irony, LinkedIn—the go-to social media site for white-collar professionals—announced that it is laying off 960 workers.

In an open memo from LinkedIn CEO Ryan Roslansky to his employees, published on the platform, he wrote, “LinkedIn is not immune to the effects of the global pandemic. Our Talent Solutions business continues to be impacted as fewer companies, including ours, need to hire at the same volume they did previously.”    ... ' 

3D Latex Printing

Nteo the use use of an embedded 3D printing system,another example of added sensor applications.   Potential healthcare applications.

Innovations Lead to 3D-Printed Latex Rubber Breakthrough
Virginia Tech News
July 13, 2020

Researchers at Virginia Polytechnic Institute and State University (Virginia Tech) have created a process to three-dimensionally (3D) print latex rubber, which could pave the way for printing elastic materials in complex geometric shapes for applications such as soft robotics, medical devices, and shock absorbers. The researchers chemically modified liquid latexes to make them printable, and built a custom 3D printer with an embedded computer vision system. Because liquid latex is extremely fragile and difficult to alter, the researchers built a scaffold around the latex particles to keep the structure in place, allowing for the addition of photoinitiators and other compounds to enable 3D printing with ultraviolet (UV) light. A camera embedded in the printer allows the machine to see the interaction of UV light on the latex resin surface and automatically adjusts printing parameters to reduce resin scattering. ... " 

Tuesday, July 21, 2020

NVIDIA and U of Florida

An Alma of mine makes an interesting AI move with NVidia.

UF announces $70 million artificial intelligence partnership with NVIDIA  
Artist's rendering of University of Florida's new AI supercomputer based on NVIDIA DGX SuperPOD architecture. 

“With AI holding the potential to revolutionize education and research – and indeed every sector of society – the University of Florida will work to ensure diversity, equity, and inclusion are at the center of this transformational change. We’re thrilled UF is demonstrating such national leadership and placing community engagement and training a 21st Century workforce at the heart of its mission.”

Peter McPherson, President, Association of Public and Land-grant Universities
The University of Florida today announced a public-private partnership with NVIDIA that will catapult UF’s research strength to address some of the world’s most formidable challenges, create unprecedented access to AI training and tools for underrepresented communities, and build momentum for transforming the future of the workforce.

The initiative is anchored by a $50 million gift -- $25 million from UF alumnus Chris Malachowsky and $25 million in hardware, software, training and services from NVIDIA, the Silicon Valley-based technology company he cofounded and a world leader in AI and accelerated computing.

Along with an additional $20 million investment from UF, the initiative will create an AI-centric data center that houses the world’s fastest AI supercomputer in higher education. Working closely with NVIDIA, UF will boost the capabilities of its existing supercomputer, HiPerGator, with the recently announced NVIDIA DGX SuperPOD™ architecture. This will give faculty and students within and beyond UF the tools to apply AI across a multitude of areas to improve lives, bolster industry, and create economic growth across the state.

“This incredible gift from Chris and NVIDIA will propel the state of Florida to new heights as it strives to be an economic powerhouse, an unrivaled leader in job creation and an international model of 21st-century know-how,” Florida Gov. Ron DeSantis said. “Over the coming years, tens of thousands of University of Florida graduates with this unique AI-oriented background will create their futures and ours, transforming our workforce and virtually every field and every industry here in Florida and around the world.”  ..   ... '

Amazon Scout Delivery Robots Expand

Been awaiting for this, looking forward to see it operational in the suburbs, on the roads and sidewalks, among people.  My camera is ready.

Amazon is testing its Scout delivery robots in Georgia and Tennessee
It’s already testing the bots in Washington and California.
Christine Fisher, @cfisherwrites in Engadget

If you live in Atlanta, Georgia or Franklin, Tennessee, your next Amazon order might arrive in one of the company’s Scout delivery robots. Amazon began testing its cooler-sized delivery bots in Snohomish County, Washington last year. They’ve been making deliveries in the Irvine area of California, and this week they popped up in Atlanta and Franklin.

Only a handful of Amazon Scout devices will operate in each city. They’ll be accompanied by a human, travel at walking speed and make deliveries Monday through Friday, during daylight hours. Customers will place their Amazon orders as usual, and there won’t be any additional cost for Scout deliveries.

Scout has successfully navigated around objects on the sidewalk -- from dogs to refrigerators left for pickup and surfboards. We still don’t know how it will verify who is opening its storage hatch or how it will unload packages if no one is there to collect them. For now, its human assistant will take care of that.  .. " 

Ethereum is 5, and is Talking

Been impressed by the direction of Blockchain player Ethereum, which has been ambitious and planning to publish a newsletter as part of their milestone.    I ask:   Where has it come from, where is it going?  Where will it take us beyond money?   When will the smart contract rise? (They promise a post on that)   Implications for cryptography, distributed consensus, Data Identity and Secure Shared intelligence?   Am Following. You can sign up for the limited newsletter at the link.

Ethereum Turns Five Next Week and We’re Producing a Special Series
By  Elaine Ramirez from Coindesk

Five years ago, a wildly ambitious project went live. Its creators envisioned a “world computer” that would transform not just money but a vast range of social interactions, pushing the boundaries of what could be done with cryptography and distributed consensus. Ethereum had arrived.

From its technical aspirations to unicorn memes, Ethereum is a culture on its own. It has spawned inventions – from digital cats to yield farming – previously unimagined and now faces a major overhaul – Eth 2.0 – to keep up with the market’s demands.

CoinDesk is marking the milestone with Ethereum at Five: a cross-platform series featuring special coverage, a limited-run newsletter and live-streamed discussions on Twitter. New issues and sessions launch daily from July 27-31.  .... 

Ethereum at Five Newsletter
Each morning during the event, our editorial team will publish a newsletter that covers the waterfront from Ethereum’s culture and lifestyle to innovations like decentralized apps, DAOs, decentralized finance and enterprise solutions, and closes out by asking the big questions about the hotly anticipated Ethereum 2.0.

Don’t know what any of that means? Led by editors Marc Hochstein, Elaine Ramirez, Christie Harkin and Zack Seward, our limited-run newsletter will be packed with educational content that eases the newbie into Ethereum, while offering the thoughtful deep-dives loyal CoinDesk readers know and love.

CoinDesk reporters and analysts Ian Allison, Leigh Cuen, Brady Dale, Nate DiCamillo, Will Foxley, Christine Kim and Hoa Nguyen will guide readers through the five years of Ethereum’s evolution, including Ethereum’s efforts to upend the banking industry, its challenges in scalability and enterprise adoption, its Woodstock moment and its status as a global lifestyle brand.

Meanwhile, we’ll provide bite-sized lessons to ease readers into the concepts of smart contracts, enterprise blockchain, yield farming and more and offer recommended reading from the CoinDesk vault.  ... "

Microsoft Studies Remote Working

Good to see this being studied in the new context. 

Microsoft Analyzed Data on Its Newly Remote Workforce
By Harvard Business Review
July 20, 2020

Microsoft employees found new touchpoints ranging from group lunches to happy hours with themes such as "pajama day" and "meet my pet." ...

Microsoft began a study of remote work four months ago when the pandemic prompted work-from-home practices. "We wanted to study how flexible and adaptable [work] might or might not be, how collaboration and networks morph in remote settings, what agility looks like in different spaces," say members of the company's Workplace Insights, Workplace Analytics, and workplace intelligence teams.

The experiment measured how work patterns across the groups were changing, and included anonymous sentiment surveys. "We looked weekly at areas such as work-life balance and collaboration by analyzing aggregated, de-identified email, calendar, and IM metadata; comparing it with metadata from a prior time period; and inviting colleagues to share their thoughts and feelings," the researchers say.

Among their findings:

Workdays are lengthening. Employees said they were carving out pockets of personal time to care for children, grab some fresh air or exercise, and walk the dog. To accommodate these breaks, people were likely signing into work earlier and signing off later.

Meetings are getting shorter. The total time for meetings each week increased by 10% overall, and individual meetings actually shrank in duration. Workers had 22% more meetings of 30 minutes or less and 11% fewer meetings of more than one hour.  ... "

Via the HBR and ACM 

KFC Testing 3D BioPrinted Chicken

One of those things that didn't even possible not too long ago.  Now being tested by a major player.

KFC to 3D Print Chicken Using Lab-Grown 'Meat of the Future'
The fast food chain wants to develop the world's first laboratory-produced chicken nuggets.

Stephanie Mlot
"Our experiment in testing 3D bioprinting technology to create chicken products can also help address several looming global problems," Raisa Polyakova, general manager of KFC Russia, said in a statement. "We are glad to contribute to its development and are working to make it available to thousands of people in Russia and, if possible, around the world."

A final product should be ready for testing this fall in Moscow, where folks are working on additive bioprinting technology that uses chicken cells and plant material to reproduce the taste and texture of meat, "almost" without involving animals. Biomeat has the same microelements of the original product without any additives (typically used in the production, processing, treatment, packaging, transportation, or storage), making it cleaner and more ethical, considering the process does not harm animals.  ... " 

Robots Given Human Like Perception of Environment

In general,  there is need for general an understanding of context, starting with physical environment.

MIT researchers have developed a representation of spatial perception for robots that is modeled after the way humans perceive and navigate the world. The key component of the team’s new model is Kimera, an open-source library that the team previously developed to simultaneously construct a 3D geometric model of an environment. Kimera builds a dense 3D semantic mesh of an environment and can track humans in the environment. The figure shows a multi-frame action sequence of a human moving in the scene.  (Videos at the link)  Paper:  https://roboticsconference.org/program/papers/79/ 

“Alexa, go to the kitchen and fetch me a snack”
New model aims to give robots human-like perception of their physical environments.
Jennifer Chu | MIT News Office

Wouldn’t we all appreciate a little help around the house, especially if that help came in the form of a smart, adaptable, uncomplaining robot? Sure, there are the one-trick Roombas of the appliance world. But MIT engineers are envisioning robots more like home helpers, able to follow high-level, Alexa-type commands, such as “Go to the kitchen and fetch me a coffee cup.”  ... 

To carry out such high-level tasks, researchers believe robots will have to be able to perceive their physical environment as humans do. 

“In order to make any decision in the world, you need to have a mental model of the environment around you,” says Luca Carlone, assistant professor of aeronautics and astronautics at MIT. “This is something so effortless for humans. But for robots it’s a painfully hard problem, where it’s about transforming pixel values that they see through a camera, into an understanding of the world.”

Now Carlone and his students have developed a representation of spatial perception for robots that is modeled after the way humans perceive and navigate the world. 

The new model, which they call 3D Dynamic Scene Graphs, enables a robot to quickly generate a 3D map of its surroundings that also includes objects and their semantic labels (a chair versus a table, for instance), as well as people, rooms, walls, and other structures that the robot is likely seeing in its environment. 

The model also allows the robot to extract relevant information from the 3D map, to query the location of objects and rooms, or the movement of people in its path.  ... " 

Offline Learning

Some good points made about using particular machine learning methods off-line.    The point that offline could be dangerous.

D4RL: Building Better Benchmarks for Offline Reinforcement Learning
By Justin Fu   By Berkeley

In the last decade, one of the biggest drivers for success in machine learning has arguably been the rise of high-capacity models such as neural networks along with large datasets such as ImageNet to produce accurate models. While we have seen deep neural networks being applied to success in reinforcement learning (RL) in domains such as robotics, poker, board games, and team-based video games, a significant barrier to getting these methods working on real-world problems is the difficulty of large-scale online data collection.

Not only is online data collection time-consuming and expensive, it can also be dangerous in safety-critical domains such as driving or healthcare. For example, it would be unreasonable to allow reinforcement learning agents to explore, make mistakes, and learn while controlling an autonomous vehicle or treating patients in a hospital. This makes learning from pre-collected experience enticing, and we are fortunate in that many of these domains, there already exist large datasets for applications such as self-driving cars, healthcare, or robotics. Therefore, the ability for RL algorithms to learn offline from these datasets (a setting referred to as offline or batch RL) has an enormous potential impact in shaping the way we build machine learning systems for the future.  ... "

Monday, July 20, 2020

Assistant IOT Regulation

Specifics of the IOT aspects would seem to be beyond these, and to any assistant, depending on the definition of an IOT.

Siri, Alexa Targeted as E.U. Probes Internet of Things
By Bloomberg     July 20, 2020

The European Commission is mounting an antitrust investigation into the Internet of Things, targeting voice assistants such as Apple's Siri and Amazon's Alexa.

The probe concerns how Silicon Valley uses data to dominate growing markets, and European Union (E.U.) Competition Commissioner Margrethe Vestager said voice assistants are central because they control how users interact with things.

Such products give technology firms access to sensitive data about consumers at a time when they continue to collect information, worrying regulators this could quash competition.

The E.U. said companies favoring their own products or setting strict terms on industry standards could impel dominant digital ecosystems and gatekeepers. For example, voice requests to buy products could circumvent competitors by immediately directing purchases to a single shopping website, like Amazon.

From Bloomberg
View Full Article    ... 

Automatically Generating Reinforcement Learning Algorithms

Can be expected that many current machine learning techniques will move towards automation.  Papers mentioned are worth looking at.

DeepMind’s AI automatically generates reinforcement learning algorithms
Kyle Wiggers in VentureBeat

 In a study  published on the preprint server Arxiv.org, DeepMind researchers describe a reinforcement learning  algorithm-generating technique that discovers what to predict and how to learn it by interacting with environments. They claim the generated algorithms perform well on a range of challenging Atari video games, achieving “non-trivial” performance indicative of the technique’s generalizability.

Reinforcement learning algorithms — algorithms that enable software agents to learn in environments by trial and error using feedback — update an agent’s parameters according to one of several rules. These rules are usually discovered through years of research, and automating their discovery from data could lead to more efficient algorithms, or algorithms better adapted to specific environments.  ... " 

Pandemic Shopping Habits

An interesting peek into corporate bahavior.

Our pandemic shopping habits are here to stay. Brands are racing to adapt  By Hanna Ziady CNN Business 

Three days a week at 7:00 am, senior Procter & Gamble executives check in with each other about their customers: what they're buying, how their needs are changing and whether the company's products are hitting the mark.  .. "

The Industrial Internet Consortium

Was just reminded of the Industrial Internet Consortium. Has some interesting documents within.

THE INDUSTRIAL INTERNET CONSORTIUM: A GLOBAL NOT-FOR-PROFIT PARTNERSHIP OF INDUSTRY, GOVERNMENT AND ACADEMIA

The Industrial Internet Consortium was founded in March 2014 to bring together the organizations and technologies necessary to accelerate the growth of the industrial internet by identifying, assembling, testing and promoting best practices. Members work collaboratively to speed the commercial use of advanced technologies. Membership includes small and large technology innovators, vertical market leaders, researchers, universities and government organizations.

Through multiple activities and programs, the Industrial Internet Consortium helps technology users, vendors, system integrators and researchers achieve tangible results as they seek to digitally transform across the enterprise. The resources of the Industrial Internet Consortium – developed collaboratively over the years by industry experts from around the globe and across all industries – give organizations the guidance needed to strategically apply digital technologies and achieve digital transformation..... "

A Look at Transfer Learning

Good generalized look at the concept of Transfer Learnig

Everything you need to know about transfer learning in AI   in TNW

Today, artificial intelligence programs can recognize faces and objects in photos and videos, transcribe audio in real-time, detect cancer in x-ray scans years in advance, and compete with humans in some of the most complicated games.

Until a few years ago, all these challenges were either thought insurmountable, decades away, or were being solved with sub-optimal results. But advances in neural networks and deep learning, a branch of AI that has become very popular in the past few years, has helped computers solve these and many other complicated problems.

Unfortunately, when created from scratch, deep learning models require access to vast amounts of data and compute resources. This is a luxury that many can’t afford. Moreover, it takes a long time to train deep learning models to perform tasks, which is not suitable for use cases that have a short time budget.

Fortunately, transfer learning, the discipline of using the knowledge gained from one trained AI model to another, can help solve these problems.

The cost of training deep learning models
Deep learning is a subset of machine learning, the science of developing AI through training examples. The concepts and science behind deep learning and neural networks is as old as the term “artificial intelligence” itself. But until recent years, they had been largely dismissed by the AI community for being inefficient.

The availability of vast amounts of data and compute resources in the past few years have pushed neural networks into the limelight and made it possible to develop deep learning algorithms that can solve real world problems.

To train a deep learning model, you basically must feed a neural network with lots of annotated examples. These examples can be things such as labeled images of objects or mammograms scans of patients with their eventual outcomes. The neural network will carefully analyze and compare the images and develop mathematical models that represent the recurring patterns between images of a similar category.

[Read: Weird AI illustrates why algorithms still need people]

There already exists several large open-source datasets such as ImageNet, a database of more than 14 million images labeled in 22,000 categories, and MNIST, a dataset of 60,000 handwritten digits. AI engineers can use these sources to train their deep learning models.

However, training deep learning models also requires access to very strong computing resources. Developers usually use clusters of CPUs, GPUs or specialized hardware such as Google’s Tensor Processors (TPUs) to train neural networks in a time-efficient way. The costs of purchasing or renting such resources can be beyond the budget of individual developers or small organizations. Also, for many problems, there aren’t enough examples to train robust AI models.

Transfer learning makes deep learning training much less demanding
Say an AI engineer wants to create an image classifier neural network to solve a specific problem. Instead of gathering thousands and millions of images, the engineer can use one of the publicly available datasets such as ImageNet and enhance it with domain-specific photos.

But the AI engineer must still rent pay a hefty sum to rent the compute resources necessary to run those millions of images through the neural network. This is where transfer learning comes into play. Transfer learning is the process of creating new AI models by fine-tuning previously trained neural networks.  ... "