Data Science newsletter – March 4, 2022

Which companies deploy machine intelligence (MI) and data analytics successfully for manufacturing and operations? Why are those leading adopters so far ahead — and what can others learn from them?

MIT Machine Intelligence for Manufacturing and Operations (MIMO) and McKinsey and Company have the answer, revealed in a first-of-its-kind Harvard Business Review article. The piece chronicles how MIMO and McKinsey partnered for a sweeping 100-company survey to explain how high-performing companies successfully wield machine learning technologies (and where others could improve).

Created by the MIT Leaders for Global Operations (LGO) program, MIMO is a research and educational program designed to boost industrial competitiveness by accelerating machine intelligence’s deployment and understanding. The goal is to “find the shortest path from data to impact,” says managing director Bruce Lawler SM ’92.

Cornell has been awarded a three-year, $8.9 million cooperative agreement from the Air Force Research Laboratory (AFRL) to launch a regional research hub that will spark collaborations between partners in academia, government and industry from across the Middle Atlantic region.

The hub’s goal is to accelerate the translation of ideas into economic impact and jobs through intellectual property and new business opportunities that benefit both commercial and defense markets.

The initial partners of the Mid Atlantic AFRL Regional Hub include Princeton University, Binghamton University, SUNY Polytechnic Institute, the U.S. Department of Energy’s Brookhaven National Laboratory and the Griffiss Institute’s Innovare Advancement Center, as well as corporate partners such as GE Research, IBM, Google, Microsoft, Corning and Amazon.

The goal of Prof. Schmidt, Dr. Zhu and their team members was therefore to design a battery significantly less than one square millimeter across and integrable on a chip, which still has a minimum energy density of 100 microwatt hours per square centimeter.

To achieve this, the team winded up current collectors and electrode strips at the microscale – a similar process also used by Tesla on the large scale to manufacture the batteries for its e-cars.

The researchers use the so-called “Swiss-roll” or “micro origami” process. A layered system with inherent tension is created by consecutively coating thin layers of polymeric, metallic and dielectric materials onto a wafer surface. The mechanical tension is released by peeling off the thin layers which then automatically snap back to roll up into a Swiss-Roll architecture. Thus, no external forces are needed to create such a self-wound cylinder micro-battery. The method is compatible with established chip manufacturing technologies and capable of producing high throughput micro-batteries on a wafer surface.

The USC project, Wearables for Health and Disease Knowledge, creates a database toolkit, titled W4H, that allows doctors to securely store and analyze medical data collected from wearable devices.

“This is AI and data science used for precision medicine,” said Cyrus Shahabi, the project’s principal investigator and chair of USC Viterbi’s Department of Computer Science.

Wearables for Health and Disease Knowledge recently received a $1.2 million grant from the National Institutes of Health that will span four years.

Shahabi says the project first came to life through the USC Integrated Media Systems Center (IMSC) and the USC Viterbi informatics program where he serves as the center’s director.

All of the materials, created through the Social and Ethical Responsibilities of Computing (SERC), a cross-cutting initiative of the MIT Schwarzman College of Computing, are now freely available via MIT OpenCourseWare (OCW). The collection includes original active learning projects, homework assignments, in-class demonstrations, and other resources and tools found useful in education at MIT.

“We’re delighted to partner with OCW to make these materials widely available. By doing so, our goal is to enable instructors to incorporate them into their courses so that students can gain hands-on practice and training in SERC,” says Julie Shah, associate dean of SERC and professor of aeronautics and astronautics.

The Carl Zeiss Foundation is funding an interdisciplinary research project which aims to establish a research center for machine learning at Johannes Gutenberg University Mainz (JGU). The TOPML: Trading Off Non-Functional Properties of Machine Learning project will be receiving support of about EUR 5 million in the foundation’s funding program on artificial intelligence. “We will investigate what is considered trustworthy AI – artificial intelligence – at a much more precise level than ever before. We not only intend to look closely at main properties, but also find out how trustworthy AI properties interact with one another and how to analyze and resolve potential trade-offs,” said Professor Stefan Kramer of the JGU Institute of Computer Science, the spokesperson for the project. In its Artificial Intelligence program, the Carl Zeiss Foundation has decided to fund a total of six research projects at the universities in Freiburg, Heidelberg, Jena, Kaiserslautern, Mainz, and Tübingen. The projects were selected by an international panel of AI experts in a two-stage evaluation process.

The University of New Mexico’s Smart Management of Infrastructure Laboratory (SMILab) has created a remote-controlled (RC) robot designed to detect potential rockslide danger. The robot, affectionately nicknamed “Brutus,” pinpoints damaged and unstable rocks for roadside safety inspectors, keeping them safely out of harm’s way during inspections.

“We have created an automated system to decrease the risk and safety hazards for inspectors,” said recent Ph.D. graduate Roya Nasimi. “With Brutus, the inspectors can remotely collect their data and their results without approaching the structure they are inspecting,” she concludes.

To accomplish this, Brutus is constructed of a mechanized rock tapping device complete with microphone and sound collecting system all mounted on an RC car base. A software algorithm written by Nasimi is incorporated to analyze the sounds created when the tapping mechanism strikes the rocks. The algorithm then calculates where instabilities exist. By studying the algorithm’s output, inspectors can find rocks that are prone to falling and remove them, proactively reducing the risk of a future rockslide.

Disney Advertising Sales on Wednesday announced that it plans to integrate new cross-channel measurement solutions into its clean room solution by way of a new partnership with Horizon Media and a proprietary identity graph. The changes reflect Disney Advertising’s larger vision of equipping advertisers with the tools to succeed in a privacy-centric, first-party-data-based future.

Walmart is rolling out Choose My Model, a high-tech feature that allows shoppers to pick a person who resembles their height, shape and skin tone to see how a clothing item would fit.

It is the first implementation of Zeekit, a start-up that Walmart acquired in May.

Facebook tried to return to its roots with the 2020 launch of Campus, a niche designed for college students. But the pilot for this private section of the social network wasn’t a success and will shut down March 10, the company confirmed to CNET.

All Campus profiles, groups, posts, events and other content will be permanently deleted and people who want to download their Campus data must do so before March 10, the company said.

The Department of Medicine at Cedars-Sinai has established a new division to explore how artificial intelligence (Ai) can be applied to help solve complex medical problems and enhance clinical care.

The new division, Artificial Intelligence in Medicine, or AIM, is led by Sumeet Chugh, MD, an associate director of the Smidt Heart Institute and a prominent expert in sudden cardiac arrest, who has long relied on technology and population-wide data to help demystify who might be susceptible to the usually fatal heart rhythm disturbance.

“Through the use of applied artificial intelligence, we can solve existing gaps in mechanisms, diagnostics and therapeutics of major human disease conditions which afflict large populations,” said Paul Noble, MD, professor of Medicine and Chair of the Department of Medicine, who created the new division. “The future of medicine lies in decoding enormous amounts of phenotypic and genotypic patient data.”

Convergent processing of the world may be a factor that contributes to social connectedness. We use neuroimaging and network analysis to investigate the association between the social-network position (as measured by in-degree centrality) of first-year university students and their neural similarity while watching naturalistic audio-visual stimuli (specifically, videos). There were 119 students in the social-network study; 63 of them participated in the neuroimaging study. We show that more central individuals had similar neural responses to their peers and to each other in brain regions that are associated with high-level interpretations and social cognition (e.g., in the default mode network), whereas less-central individuals exhibited more variable responses. Self-reported enjoyment of and interest in stimuli followed a similar pattern, but accounting for these data did not change our main results. These findings show that neural processing of external stimuli is similar in highly-central individuals but is idiosyncratic in less-central individuals. [full text]

“Algorithms reinforce human biases. They amplify them. So algorithmic biases can be much worse than human biases and can have greater impacts,” said Cecilia Aragon (pictured), professor of human-centered design and engineering at the University of Washington.

Aragon spoke with Lisa Martin, host of theCUBE, SiliconANGLE Media’s livestreaming studio, in advance of the Women in Data Science (WiDS) Worldwide Conference 2022. They discussed the importance of recognizing the human aspect of data science and how Aragon overcame her early “imposter syndrome” to become a data scientist, college professor, aerobatic pilot and author.

A group of researchers at Georgia Tech are looking to better understand financial inclusion, and they told Hypepotamus it could help propel the next generation of responsible-focused FinTech startups in the city.

The Financial Services Innovation Lab, housed inside Georgia Tech’s Scheller College of Business, has been a hub for finance research on campus and across the Metro region for some time. The Lab just announced a new partnership with Atlanta-based Equifax to help better understand what it means to make FinTech and the broader financial sector more inclusive.

Director Sudheer Chava told Hypepotamus that responsible finance, machine learning, analytical finance, FinTech, and cryptocurrencies are the main research areas for the lab. Within responsible and sustainable finance, the Lab is tapping into large, anonymous datasets to better understand what things really serve as “barriers to credit access.” The partnership will give the Lab comprehensive alternative data sources and access to credit modeling experts through the Equifax Data Science Lab.

UT Southwestern Medical Center is pleased to announce a $100 million gift from the O’Donnell Foundation to endow and support its new School of Public Health. This investment is the largest gift to a School of Public Health at a public university in the U.S. and matches the third largest gift supporting any School of Public Health.

The O’Donnell Foundation, established by visionary philanthropist Peter O’Donnell Jr., who died last year, and his late wife, Edith, made this gift for its unprecedented potential to accelerate the momentum of the recently established School. In recognition of this gift, UT Southwestern has named the new school the Peter O’Donnell Jr. School of Public Health. It is the first new school established at UT Southwestern in more than 50 years. It joins UT Southwestern’s Medical School, School of Health Professions, and Graduate School of Biomedical Sciences.