Machine Learning Determines Research Impact

 Good use example made here, note training made on key metadata.  Relatively straight forward.

Using ML to Predict High-Impact Research

MIT News, Becky Ham May 17, 2021

Researchers at the Massachusetts Institute of Technology (MIT) have developed an artificial intelligence framework able to predict future high-impact technologies based on patterns found in published scientific studies. The framework, DELPHI (Dynamic Early-warning by Learning to Predict High Impact), identified all pioneering papers on a list of key foundational biotechnologies as early as the first year after their publication. MIT's James W. Weis said the framework "functions by learning patterns from the history of science, and then pattern-matching on new publications to find early signals of high impact." DELPHI, which was trained on a full time-series network of journal article metadata, identified more than twice as many high-impact papers compared to citation numbers alone.  ... ' 

What Google is Emphasizing in AI

Good overview of areas Google is working and emphasizing from Google I/O.  Instructive.  Details at the link.  Not heavily technical. 

11 ways we're innovating with AI

Christine Robson, Director of Product, Google Research

AI is integral to so much of the work we do at Google. Fundamental advances in computing are helping us confront some of the greatest challenges of this century, like climate change. Meanwhile, AI is also powering updates across our products, including Search, Maps and Photos — demonstrating how machine learning can improve your life in both big and small ways. 

In case you missed it, here are some of the AI-powered updates we announced at Google I/O:.... '

Timeline for Quantum Computing

 Monte Carlo methods are where speed is needed.  And in theory Quantum Computing could get us there.   Still some nagging issues exist. 

The Timeline for Quantum Computing is Getting Shorter  By NetworkWorld, May 19, 2021

Financial traders rely heavily on computer financial simulations for making buying and selling decisions. Specifically, "Monte Carlo" simulations are used to assess risk and simulate prices for a wide range of financial instruments. These simulations also can be used in corporate finance and for portfolio management.

But in a digital world where other industries routinely leverage real-time data, financial traders are working with the digital equivalent of the Pony Express. That's because Monte Carlo simulations involve such an insanely large number of complex calculations that they consume more time and computational resources than a 14-team, two-quarterback online fantasy football league with Superflex position.

Consequently, financial calculations using Monte Carlo methods typically are made once a day. While that might be fine in the relatively tranquil bond market, traders trying to navigate more volatile markets are at a disadvantage because they must rely on old data. If only there were a way to accelerate Monte Carlo simulations for the benefit of our lamentably ladened financial traders!

Soon there will be, according to financial services giant Goldman Sachs and QC Ware, a quantum-as-a-service provider that develops applications to run on near-term quantum-computing hardware. Researchers for the two partners reportedly have designed new quantum algorithms for running Monte Carlo simulations on near-term quantum hardware expected to be available in five to 10 years.

From NetworkWorld

AI Teaching Itself

 Its al about setting goals, metagoals in the right context. 

Why AI That Teaches Itself to Achieve a Goal Is the Next Big Thing

by 7wData  April 24, 2021  by Kathryn Hume and Matthew E. Taylor

April 21, 2021 HBR

Summary.   What’s the difference between the creative power of game-playing AIs and the predictive AIs most companies seem to use? How they learn. The AIs that thrive at games like Go, creating never before seen strategies, use an approach called reinforcement... 

Lee Sedol, a world-class Go Champion, was flummoxed by the 37th move Deepmind’s AlphaGo made in the second match of the famous 2016 series. So flummoxed that it took him nearly 15 minutes to formulate a response. The move was strange to other experienced Go players as well, with one commentator suggesting it was a mistake. In fact, it was a canonical example of an artificial intelligence algorithm learning something that seemed to go beyond just pattern recognition in data — learning something strategic and even creative. Indeed, beyond just feeding the algorithm past examples of Go champions playing games, Deepmind developers trained AlphaGo by having it play many millions of matches against itself. During these matches, the system had the chance to explore new moves and strategies, and then evaluate if they improved performance. Through all this trial and error, it discovered a way to play the game that surprised even the best players in the world.

If this kind of AI with creative capabilities seems different than the chatbots and predictive models most businesses end up with when they apply machine learning, that’s because it is. Instead of machine learning that uses historical data to generate predictions, game-playing systems like AlphaGo use reinforcement learning — a mature machine learning technology that’s good at optimizing tasks. To do so, an agent takes a series of actions over time, and each action is informed by the outcome of the previous ones. Put simply, it works by trying different approaches and latching onto — reinforcing — the ones that seem to work better than the others. With enough trials, you can reinforce your way to beating your current best approach and discover a new best way to accomplish your task.

Despite its demonstrated usefulness, however, reinforcement learning is mostly used in academia and niche areas like video games and robotics. Companies such as Netflix, Spotify, and Google have started using it, but most businesses lag behind. Yet opportunities are everywhere. In fact, any time you have to make decisions in sequence — what AI practitioners call sequential decision tasks — there a chance to deploy reinforcement learning.

Consider the many real-world problems that require deciding how to act over time, where there is something to maximize (or minimize), and where you’re never explicitly given the correct solution. For example: ....   "

How Effective are top Education Apps?

 Good feedback as to what works in the new contexts.

Top Educational Apps for Children Might Not Be as Beneficial as Promised

Penn State News, Katie Bohn, May 11, 2021

An analysis of the most frequently downloaded educational apps for kids by a team of researchers led by the Pennsylvania State University Brandywine found such apps may not provide high-quality educational experiences. The researchers used previous research on the pillars of learning to develop criteria for the assessment of the top 100 children's educational apps from the Google Play and Apple apps stores, among others. After the apps were scored from 0 (low) to 3 (high) for each pillar of learning, the researchers found a score of 1 was most common for each app with regard to all four pillars. The University of Michigan’s Marisa Meyer said, “If app designers intend to engender and advertise educational gains through use of their apps, we recommend collaborating with child development experts in order to develop apps rooted in the ways children learn most effectively. .... '

P&G Admits What Worked, What Didn't in Pandemic

Some rare inside information re Digital details from my former colleagues at P&G.  Recall I have mentioned here that P&G was an early pioneer in AI based methods.  Its good you can admit to what did not work.  Lets you adjust your directions and methods in the future. 

P&G's Benjamin Spiegel Admits What He Got Wrong

The global chief digital officer also discussed AI, 'maskne,' virtual influencers and the lipstick effect.

By Paul Hiebert   in AdWeek

Benjamin Spiegel, global chief digital officer for Procter & Gamble’s beauty segment, is not afraid to admit that some changes in consumer behavior he thought might occur during the pandemic … well, did not occur.

“There were a lot of things that we assumed would happen early on that, in the end, didn’t quite pan out,” noted Spiegel while speaking at Adweek’s Elevate: AI summit.

One was an expected boom in direct-to-consumer shopping. Instead, Spiegel explained, retailers got creative and have updated their click-and-collect and last-mile delivery options. “That was a big one for me that I got totally wrong,” he said.

Another prediction was that people would struggle with working at home—that the ideas, creativity and collaboration just wouldn’t be there. While remote work certainly has its challenges, Spiegel said people have shown resilience and embraced new technology that enables workers to get the job done.

Spiegel said the company has begun thinking about campaigns as “always on,” as opposed to initiatives that start and stop. “The time[s] of day where people make decisions [and] consume content have definitely changed,” he said.

P&G has also tracked the popularity of certain search terms, such as “maskne” (i.e. acne caused by wearing a face mask), and responded with tailored messaging and educational content. Usually, trends come and go, Spiegel said, but trends during Covid-19 had a “big impact on how people thought about product and usage.”

The lipstick effect—the theory that shoppers still spend money on small indulgences during economic downturns—is still going strong during the pandemic, Spiegel said. For example, their research found that instead of buying a piece of clothing that costs $200, consumers would purchase a $40 beauty product.

“People went into saving money in certain places, but then they ended up spending more in other places,” he said.

For the quarter ending March 31, P&G’s beauty business, which includes the brands Olay, SK-II and Old Spice, saw net sales climb to $3.3 billion, up 9% compared to the same period in 2020.  ... '

Boxes with Personality

 Does personality buy you something with speaking devices?   With Robots?   Maybe, sometimes, in the right environment.  Its been tried many times.   Alexa does a bit, but not enough. A stronger conversational flow, with active memory would help.  Clear value goals.  Here an automated talking box on wheels from Norway makes the claim, for in hospital systems.  

An Automated Box on Wheels—with Personality   By Norwegian SciTech News   May 17, 2021

People find boxy talking robots humorous and engaging, which may make robotic technology more acceptable in everyday life, according to new research.

Norwegian University of Science and Technology (NTNU) scientists studied robotic Automated Guided Vehicles (AGVs) deployed at St. Olavs Hospital in Trondheim, Norway, which were equipped with voices with a distinctive local dialect, rather than a generic Norwegian voice.

NTNU's Roger A. Soraa said, "We found that these robots, which were not created to be social robots, were actually given social qualities by the humans relating to them."

Soraa said this quality helps people accept the robots, which he described as "emissaries of technologies."   ... 

This robot transports trolleys around St. Olavs Hospital in Trondheim, Norway. When people hear it speak, they perceive it as a kind of funny animal-like creature, rather than a robot.

From Norwegian SciTech News

Controlling Drones with Minds

 Not unexpected, have seen the forerunners of the idea.  Will also require much in the way of collaboration between drones, goals, controllers and troops.  It will ultimately succeed.  

The US military is trying to read minds

A new DARPA research program is developing brain-computer interfaces that could control “swarms of drones, operating at the speed of thought.” What if it succeeds?  ... ' 

Technology Review  by Paul Tullis archive page

Malware Tricks

 Recorded Future, a podcast and text. With a look to the future of cybersecurity. This is an area where you are likely to see advanced tech applied, for good and bad.  

Malware Party Tricks and Cybersecurity Trends

APRIL 26, 2021 •  Caitlin Mattingly

This week we welcome back to our program security pioneer Graham Cluley. After starting his career writing the original version of Dr. Solomon’s Antivirus Toolkit for Windows, Graham moved on to senior positions at Sophos and McAfee. In 2011 he was inducted into the Infosecurity Europe Hall of Fame. These days, he’s an independent blogger, podcaster and media pundit.

Our conversation takes a sometimes nostalgic look back at the origins of computer malware, what it was like fighting the good fight back then, how things have developed over the years, and what he thinks the future may hold.   ... " 

Emergence of AI Factories

 Most interesting, what are the particular skills involved?   Software tools at what level.   Linkages to data?  Considerable effort if there are 60 projects underway. 

Mayo Clinic, Others Use 'AI Factories' to Speed AI Development   By The Wall Street Journal, May 18, 2021

Mayo Clinic and other organizations are using an assembly-line approach to artificial intelligence (AI) development, where small teams use a common set of software tools and procedures to speed the production of AI applications while cutting costs.

The Minnesota-based healthcare provider launched what it calls its AI factory in September and is now getting into full production, with about 60 projects under way, said James Buntrock, vice chair of the department of information technology at Mayo Clinic.

"It's [a] more consistent process to produce algorithms," Mr. Buntrock said.

One application in development aims to analyze medical images to identify and classify "biomarkers"—measurable medical signs, such as excess abdominal fat—which could help predict patient health. Mr. Buntrock declined to quantify the speed and cost savings of the AI factory effort, but said they are significant.

The main reason companies are turning to the AI factory model is to increase their success rate with AI, said Erick Brethenoux, a vice president analyst at technology research and advisory firm Gartner Inc.

From The Wall Street Journal

Cryptographers on Non Fungible Tokens

 Not quite sure why they had a direct problem with NFTs.  Besides some general level of incredulity.  I will agree.

ACM NEWS

Cryptography Experts Trash NFTs on First Day of RSA Conference  By Mashable

The annual RSA Conference brings together some of the brightest minds in cryptography to discuss advances in the field, the year's biggest hacks, and where the cybersecurity industry is heading. This year, it just so happened to kick off with a pronounced dunk on NFTs. 

The first main event of RSA 2021 was the cryptographers' panel, which followed Monday's opening keynote. At the start of the panel, Ron Rivest, a famed cryptographer who co-created RSA public-key encryption, derided non-fungible tokens as worth even less than the famed tulips of tulip mania. 

At least with actual tulips, argued Rivest, "you can own them, you can posses them, you can plant them, you can enjoy them."   NFTs, Rivest observed, aren't even like pictures of tulips. They're more akin to digital tokens that point at a picture of a tulip. 

"It's a bit like homeopathic medicine," said Rivest. "You dilute it, you dilute it, you dilute it, and you say, 'What's left?'"  ... "

Virtual Agents

 Virtual Agents, not commonly mentioned these days.   Adding 'intelligent' implies some goal, directive.  Typically at the individual agent level.    When we often built these we  constructed them at a level where they represented elements like consumers. Behaving similarly.  Driven by data. 

7 Key Industries Being Transformed by Data Savvy Virtual Agents  in SmartDataCollective

Big data has become very useful in recent years, which has led to the inception of virtual agents.

There are a number of new fields that have opened up due to recent advances in big data. Big data has played a huge role in the evolution of business development. One field that has emerged as a result of new developments in big data technology is the virtual agent.

A Better Way to Model Data

There are a number of new fields that have opened up due to recent advances in big data. Big data has played a huge role in the evolution of business development. One field that has emerged as a result of new developments in big data technology is the virtual agent.

Intelligent virtual agents (IVAs) are attractive to businesses in dozens of industries thanks to their convenience and their AI capabilities. The already growing market got a boost when many businesses went wholly virtual during the COVID-19 pandemic

Companies needed to figure out how to offer top-notch customer service while still keeping teams lean and productive. Virtual agents help create a seamless customer experience while empowering human employees to work on pressing tasks that need the human touch.

The overall IVA market is predicted to grow 34% by 2027, and it was valued at $3.7 billion in 2019. Here’s a look at some of the industries this growing technology is transforming.  ... '

Google Project Starline for Quality Personal Communications

Google has created and writes about their Project Starline.   Which they say a very high quality communications capability that mimics live face to face interaction.  Including eye movements and other emotional cues.   Without any specialized headsets or glasses like for VR.   They are testing it internally, and imply it works with 3D imaging and a 3D display in real-time.   No mention of it working on a phone.  Ultimately meant to compete with other pandemic era needs?

< ... An early technology, this "magic window" makes you feel like you're together, even when you're apart ... >

Project Starline: Feel like you're there, together    By Clay Bavor   VP, Google  in PRNewswire

People love being together — to share, collaborate and connect.  And this past year, with limited travel and increased remote work, being together has never felt more important.

Through the years, we’ve built products to help people feel more connected. We’ve simplified email with Gmail, and made it easier to share what matters with Google Photos and be more productive with Google Meet. But while there have been advances in these and other communications tools over the years, they're all a far cry from actually sitting down and talking face to face.

We looked at this as an important and unsolved problem. We asked ourselves: could we use technology to create the feeling of being together with someone, just like they're actually there?

To solve this challenge, we’ve been working for a few years on Project Starline — a technology project that combines advances in hardware and software to enable friends, families and coworkers to feel together, even when they're cities (or countries) apart.

Imagine looking through a sort of magic window, and through that window, you see another person, life-size and in three dimensions. You can talk naturally, gesture and make eye contact.

To make this experience possible, we are applying research in computer vision, machine learning, spatial audio and real-time compression. We've also developed a breakthrough light field display system that creates a sense of volume and depth that can be experienced without the need for additional glasses or headsets.

The effect is the feeling of a person sitting just across from you, like they are right there. 

A breakdown of how Project Starline works from 3D imaging to real-time data compression to 3D display.  Key breakthroughs include 3D imaging, real-time compression and a 3D display.

One of the things we are most proud of is that as soon as you sit down and start talking, the technology fades into the background, and you can focus on what's most important: the person in front of you. 

Project Starline is currently available in just a few of our offices and it relies on custom-built hardware and highly specialized equipment. We believe this is where person-to-person communication technology can and should go, and in time, our goal is to make this technology more affordable and accessible, including bringing some of these technical advancements into our suite of communication products.  .... "

Signal Analytics and Baby Care

I see some of my former employers are involved.   Platform for predictive demand forecasting based on marketing decisions is a good thing. 

Signals Analytics Launches AI-Driven Intelligence Platform for Booming Baby Care Market  in PRNewswire

Platform allows baby care brands to leverage wide dataset of consumer and market connected insights in near real time.

NEW YORK, May 18, 2021 /PRNewswire/ -- Today, Signals Analytics, an AI-powered consumer and market intelligence platform, is adding Baby Care to its rapidly growing list of industry-specific offerings. Fresh off the heels of its successful Consumer Electronics and Apparel category launches earlier this year, the Baby Care expansion comes as demand for connected, external data and predictive analytics spikes in helping support cross-functional, insight-driven decision making.

Signals Analytics' new Baby Care solution offers a view into cross-market trends as well as category-specific insights for Baby Wipes & Diapers and Baby Personal Care, with a focus on Skin Care, Hair Care, and Body Hygiene - helping brands compete in this crowded sector. Pulling from over 13,000 external data sources -- spanning Instagram, Amazon, Target, and LexisNexis -- the platform will help fuel actionable go-to-market decisions in areas such as product innovation, pipeline prioritization, brand messaging, and marketing execution, to enable brand relevance and category share growth. 

Valued at $67 billion globally in 2020, the baby care market is expected to reach $88.7 billion by 2026, in part due to a rise in disposable parental income, older parental ages, and increased consumer awareness of baby hygiene. Historically, the market has been dominated by a few powerplayers, yet smaller, high-growth challenger brands are gaining momentum and market share. Ecommerce availability has leveled the playing field giving newer, more niche baby brands previously untapped exposure. With competition heating up and new products rolling out, it is imperative that brands have insight into what will drive new parents to purchase and understand the concerns they have top of mind. ... ' 

Massive Cell Data sets

Analyzing organisms with minimal computing power. 

Algorithm Uses Online Learning for Massive Cell Datasets

Michigan Medicine, Kelly Malcolm, April 19, 2021

An algorithm developed by University of Michigan (U of M) researchers employs online learning to accelerate the analysis of enormous cell datasets, using the amount of memory found on a standard laptop computer. The algorithm enables new datasets to be added to existing ones without reprocessing the older datasets, and allows researchers to segment datasets into mini-batches so less memory is required for processing. U of M's Joshua Welch said, "Our technique allows anyone with a computer to perform analyses at the scale of an entire organism. That's really what the field is moving towards."

Grand Transitions

Makes you think.

A Theory of (Almost) Everything

A new book / Podcast  analyzes the seven grand transitions that have formed the modern world  By Steven Cherry

We’ve been speaking with Václav Smil, Distinguished Professor Emeritus at the University of Manitoba, Fellow of the Royal Society of Canada, Member of the Order of Canada, scholar extraordinaire, and author of a new must-read book:

Grand Transitions: How the Modern World Was Made.

 Steven Cherry Hi this is Steven Cherry for Radio Spectrum.

If there’s one thing we can all agree on, it’s that the world is not only changing quickly, it’s changing at a faster rate than ever. Or does it just seem that way?

Surely we can all agree that the Industrial Revolution has changed everything. Or has it? One noted economist says there in fact were three industrial revolutions, and only one of them—the second one, from about 1870 to 1914, was important. In fact he largely discounts what we call the information revolution as insubstantial.

If you wanted to study the great trends and transitions of civilization—not just Western Civilization, but all of it—and break it down into epochs, and choose from the various transitions the five or seven most significant ones, and study the interplays of these transitions—which are causes of the others, and to what degree, and why some occur quickly and others—like the electric car—are postponed for a hundred years; if you wanted to do all that, it would take a lifetime of study.

In fact, you’d have to write ten or thirty books each one of which looks at some aspect of our world from a height of 30,000 feet, and then write an eleventh or thirty-first book that was the encapsulation of all that wisdom.

That certainly seems impossible. The last true Renaissance person, someone who knew pretty much all that was known at the time, might have been Aristotle, with asterisks for Franklin and Diderot, and maybe Bertrand Russell.  ... ' 

Superforecasting Update

For some time have followed GoodJudgment  and their 'superforecasting' services.   Most recently got a new overview with considerable efforts and claims.    Since very accurate forecasting would seem to be fundamentally unbeatable in many domains, why believe it?  May depend on the definition of a correct forecast in a given context.  See their work with returns to normality with Covid.  And 'Right ways to think about the future'.

Why Superforecasting?

When you make decisions based on precise probability forecasts, rather than hunches, the benefits are game-changing.

Gain a competitive edge

Making decisions without precise probability forecasts is like playing poker without counting cards. Outsmart your competition with well-calibrated forecasts.

Manage risk

Convert strategic uncertainty into manageable risk to avoid costly mistakes.

Seize opportunities

Identify expected outcomes before others recognize them so that you can capitalize on opportunities quickly and with confidence.  From groundbreaking theory to powerhouse practice

In 2011, IARPA – the US intelligence community’s equivalent to DARPA – launched a massive competition to identify cutting-edge methods to forecast geopolitical events. Four years, 500 questions, and over a million forecasts later, the Good Judgment Project (GJP) – led by Philip Tetlock and Barbara Mellers at the University of Pennsylvania – emerged as the undisputed victor in the tournament. GJP’s forecasts were so accurate that they even outperformed intelligence analysts with access to classified data.

Good Judgment Inc is now making this winning approach to harnessing the wisdom of the crowd available for commercial use. Our clients benefit from the externally validated forecasting methodology that made the Good Judgment Project so successful.

Today, Good Judgment’s professional Superforecasters deliver unparalleled accuracy on forecasting questions across the political, economic and social spectrum. And, we train others to apply this evidence-based methodology within their own teams.  .... '

Amazon Echo Tries Smart Glasses

A number of companies have tried the wearable assistant idea, announced originally in December, now Amazon will try the idea in earnest.  Will the idea work this time.  What key aspects are important?  In some tests we did with Google version, there was concern about how people using the glasses were perceived.  Were they continually taking pictures?    Is the idea for continuous use, which Amazon appears to imagine, or just for specialized assistant tasks?  We tried in plant maintenance tasks.   What would the key skills for glasses look like? 

Amazon Echo Frames arrive in blue light and sunglasses forms for summertime  in DigitalTrends  By Patrick Hearn  May 18, 2021 

The Amazon Echo Frames originally launched in December but came with a somewhat limited series of options. Today, Amazon announced two new models of the Echo Frames: A set of polarized sunglasses and a set of blue-light filtering glasses. The new options come after a lot of customer feedback asking for different options and styles.

The sunglasses themselves are available in two different versions, too. One is a classic set of sunglasses, while the other is a set of blue mirror sunglasses. Both versions are IPX4-rated, which means they are perfect for wearing while working out or walking outside. While they shouldn’t be worn while swimming, they can withstand a few splashes of sweat or rain.

On the other hand, the Echo Frames with a blue-light filter are perfect for people that spend a lot of time at their computers. Blue light can interfere with your circadian rhythm and interrupt your sleep cycle, so the blue light filter makes it possible to play video games or browse the web in the evenings without worrying about disturbing your rest.

The polarized blue mirror sunglasses will start shipping on May 18, while the Echo Frames with blue-light filtering lenses and the polarized classic sunglasses begin shipping on June 9. Both options are available in the classic black frames for $270.

The Echo Frames are more than just glasses. They provide access to Alexa with a touch of a button, as well as give you easy access to music, podcasts, and more without the use of headphones. You can control your music and playlists by saying “Alexa, pause,” or “Alexa, next.”  ...  "

Using the Rust Language for Code Safety

In a recent dev oriented conversation someone mentioned Rust as a programing Language to improve code safety.   I heard about it again today re Facebook's usage.   Also the Rust Foundation seems to have support from major players.   Here is an overview and history of its use.  

A brief history of Rust at Facebook

Facebook is embracing Rust, one of the most loved and fastest-growing programming languages available today. In addition to bringing new talent to its Rust team, Facebook has announced that it is officially joining the nonprofit Rust Foundation. Alongside fellow members including Mozilla (the creators of Rust), AWS, Microsoft, and Google, Facebook will be working to sustain and grow the language’s open source ecosystem.

For developers, Rust offers the performance of older languages like C++ with a heavier focus on code safety. Today, there are hundreds of developers at Facebook writing millions of lines of Rust code. And while it’s clear that Facebook is increasingly invested in the future of the language, it’s important to understand how we grew to this point.  ... '

Google Machine Learning Glossary

 Ajit Jaokar ( Course Director: Artificial Intelligence: Cloud and Edge Implementations - University of Oxford  )      Sends along his AI Newsletter, he writes: 

Open this article on LinkedIn to see what people are saying about this topic. Open on LinkedIn

Edition 4: A simplified glossary for machine learning and deep learning

Our course at the #universityofoxford on developing AI starts in the first week of June and registrations close next week.   

In this post, I am going to share something I am working on for this course. I created the attached glossary as a mechanism to teach AI (machine learning and deep learning) because I could not find a concise machine learning glossary for getting started in AI.

The glossary is based on the excellent Google Machine Learning Glossary:   

Here is the updated ML Glossary.   ...   From Feynlabs.AI  ....   Very good, a long PDF with considerable detail....