Data Science newsletter – November 22, 2021

Before there was arXiv, there was Joanne Cohn. In the late 1980s, she was a postdoc at the Institute for Advanced Study in Princeton, New Jersey, working in the heady, fast-paced field of string theory. She started an informal exchange of string theory manuscripts that eventually became the arXiv preprint server, which has since revolutionized the way scientists share ideas and announce findings. … Today Cohn is a research scientist in theoretical cosmology at the University of California, Berkeley. Here’s the story she told Physics Today, edited for brevity and clarity, about her pivotal preprint-sharing project.

Various macroscopic analogies that replicate multi-body interaction and specific geometry of the [n-body] problem are tremendously insightful. Among those, now, a simple table-top experimental method developed by Stanford University engineers. All that’s needed to begin is a slippery surface (say a glass slide), a permanent marker and a mixture of water and propylene glycol, a common ingredient in food coloring.

With these supplies, the researchers invented a new way to rapidly prototype complex geometries mirroring symmetries present in problems of interest. Instead of planets strewn about the solar system, many tiny droplets interact with each other at a distance and the observer can directly watch and manipulate how the system evolves over time.

The newly launched Synthesis and Processing Informed by Rational Algorithmic Learning (SPIRAL) Center, one of 10 projects receiving Department of Energy funding, aims to use artificial intelligence to optimize the creation of materials needed for solar power.

The News-Letter spoke to SPIRAL’s Hopkins group leaders Paulette Clancy, department head of chemical and biomolecular engineering, and Rigoberto Hernandez, Gompf family professor of chemistry, about the foundation and future directions of the project.

Like America’s hyper-rich, the nation’s elite universities are amassing vast wealth with astonishing speed — and raising some uncomfortable questions in the process.

Amid the running debates over the cost and value of a college education today, two key issues loom large. The first: To what extent should rich institutions like Harvard, with sticker prices approaching $80,000 a year, share their growing wealth with students? And should they — the Elon Musks and Jeff Bezoses of higher education — pay more in taxes?

With their newfound wealth, schools are bolstering financial aid and seeking to improve affordability for low-income families. They’re increasing faculty and staff benefits and also using some money to offset the financial hit from the Covid-19 pandemic.

A UCI interdisciplinary research team associated with the Skin Biology Resource Center has received a three-year, $2 million grant from the Chan Zuckerberg Initiative to generate a comprehensive skin cell atlas that will help explain ancestral differences in skin biology and disease. The project is part of a worldwide effort to create the Human Cell Atlas, which will map every cell type in the human body and serve as reference data broadly representative of race, ethnicity, ancestry and other determinants of health

Because of the emphasis on data know-how, when companies look for citizen data science candidates, it makes sense to review the ranks of disciplines within the company (e.g., engineering, software development, data analysis) where some of these skills may already be resident. For example, a data analyst in IT or a software developer is likely to grasp the idea of modeling data for analysis easier than an individual in purchasing, marketing or accounting.

There is also an added bonus if companies can cross train employees who have business acumen and strong relationships with end users throughout the company. These people might have more ability to understand and dissect the information needs of the business than a data scientist who knows math, statistics and analytics—but who has no working knowledge of the business.

Can citizen data scientists supplant data scientists altogether? No. But they can ease the load on the data science staff.

The Defense Innovation Unit released its initial “Responsible AI Guidelines” document Nov. 15, with intent to operationalize the Defense Department’s ethical principles of artificial intelligence into its commercial prototyping and acquisition efforts.

“DIU’s RAI guidelines provide a step-by-step framework for AI companies, DOD stakeholders and program managers that can help to ensure that AI programs are built with the principles of fairness, accountability and transparency at each step in the development cycle of an AI system,” Jared Dunnmon, PhD, technical director of the artificial intelligence/machine learning portfolio at DIU said.

One particularly concerning reality of contemporary clinical care is the expanding mental health crisis sweeping the country in the wake of the pandemic.

The impact on children, in particular, was emphasized poignantly by Zak Kohane, a pediatric endocrinologist by training, who now chairs Harvard’s Department of Biomedical Informatics. 

“There is an unbelievable disaster happening in pediatric mental health right now,” Kohane explains.  “Our local hospital [i.e. Boston Children’s Hospital] is not accepting elective admissions — not because of COVID, not because of trauma, but because our psych wards are overflowing. And this is, by the way, not just us. It’s nationwide.”

An important challenge in addressing mental-health problems is that trends can be difficult to detect because detection relies heavily on self-disclosure. As such, helplines — telephone services that provide crisis intervention to callers seeking help — might serve as a particularly useful source of anonymized data regarding the mental health of a population. This profiling could be especially useful during the COVID-19 pandemic, given the potential emergence or exacerbation of mental-health problems1. Together, the threat of disease to oneself and others that is associated with a local epidemic2, the restrictiveness of local non-pharmaceutical interventions (such as stay-at-home orders) and the potential associated loss of income could have contributed to a decline in the mental health of a population while at the same time inhibiting or delaying people’s search for help for problems3. Writing in Nature, Brülhart et al.4 present evidence suggesting that helpline-call data can be used to monitor real-time changes in the mental health of a population — including over the course of the COVID-19 pandemic.

Despite the program’s name, I think it’s not so much about individual users repairing their own personal devices. The biggest ramification, I think, will be that the program will allow unofficial independent repair shops to procure genuine OEM Apple replacement parts and service manuals. There are tons of people around the world (including here in the U.S.) who don’t live near an Apple store or an Apple-authorized repair shop. A lot of those people, though, might live near (or at least nearer) an independent repair shop. If those repair shops can now order genuine Apple parts and manuals, that’s a win, and maybe a bigger deal than I thought yesterday.

There’s also this factor: if the device in need of repair is still usable — say, an iPhone with a cracked but functional screen, or a MacBook with one or more broken but nonessential keys — it might be a lot more appealing for a user who doesn’t live near an Apple-authorized repair shop to go to a local independent shop for same-day service than to ship their device to Apple for official service.

On the flip side, though, I think a lot of the “Apple’s repair policies are screwing people” sentiment is based on the misconception that Apple grossly overcharges for repairs.

Internet access has always been a problem for Faylene Begay, a single mother of four living in Tuba City, Ariz.

Before the pandemic, she didn’t have an internet connection at her home on the Navajo Nation Reservation — all she had was an old phone with limited data. Back then, her lack of connection was a nuisance as she worked her way through classes at Diné College.

But when her college campus closed in spring 2020, internet access became a major challenge: She could complete all of her assignments, but uploading them required a strong internet connection, which she didn’t have.

Jeannine Cavender-Bares began her scientific career with her hands in the dirt. Now, she has her eyes in the sky. As a teenager she helped her biologist dad dig deep into the leaf litter of Florida oak forests to catalog the diversity of slime molds, protists best known for uniting into unsightly blobs that creep across the landscape. But it was the oaks overhead that really fascinated her and became the focus of high school science fair entries and her graduate school research.

In forest plots across the southeastern United States, she planted acorns to study how local conditions affect photosynthesis and growth of different oak species. She froze the seedlings’ stems to study how they transport water to leaves and climbed into the forest canopy to measure gases emitted by mature trees.

But such studies could only provide a snapshot of one forest at a time. To get the big picture of forests around the world, Cavender-Bares has sought a higher vantage. Now a plant ecologist at the University of Minnesota, Twin Cities, Cavender-Bares has devised ways to translate light measured by spectrometers flown over forests into insights about their health and resilience. She and others have found this light, captured from an airplane or satellite, holds clues to intimate details such as photosynthesis levels, the genetic diversity of the trees, and even the microbial inhabitants of the soil they grow in.

Essentially, April19 [a UK company] taught the AI the work of Alexander Shulgin. Shulgin is the godfather of modern psychedelic science and most notably invented MDMA among an array of other beneficial substances he would sample with his wife before bringing them to dinner parties with friends. April19 told the AI to go nuts. In the process, the AI identified 17 new compounds that have the potential to provide healing benefits.

But it gets crazier.

With the DEA’s approval of an increase in the psychedelic research supply chain, there is going to be tons of data for the AI to process over the next 24 months as more psilocybin and MDMA is eventually divided up between researchers across the nation. The AI will be able to use any kind of human data it’s able to get its hands on moving forward to make its process even more refined.

The University of Michigan has no plans to return a $4 million donation from Activision Blizzard CEO Bobby Kotick in the wake of allegations that he hid reports of sexual misconduct and rape within his company from its board.

Kotick, who attended Michigan in the early 1980s before becoming one of the most powerful publishers in gaming, donated the money earlier this year to help the school jumpstart its esports curriculum.

Check out this article that images cell apoptosis and interprets with deep learning!

The 5th Annual Women in Data Science (WiDS) Datathon will launch in January, organized by Stanford University, Harvard University IACS, and the WiDS Datathon Committee, addressing an important way to mitigate the effects of climate change with a focus on energy efficiency. The WiDS Datathon Committee is partnering with experts from many disciplines at Climate Change AI, Lawrence Berkeley National Laboratory (Berkeley Lab), US Environmental Protection Agency (EPA), and MIT Critical Data.

“The WiDS Datathon has consistently featured a social impact challenge,” said Karen Matthys, Co-Director of the WiDS initiative at Stanford University. “This esteemed group of collaborators created a challenge related to one of the most critical issues of our time: climate change.”

A modern alternative to the core encryption technology bundled with Linux distributions is ready for testing after five years of development.

The Linux Random Number Generator (LRNG), which relies on several computing functions to act as a source of entropy, is designed to be a drop-in replacement for the long-established /dev/random function.