Data Science newsletter – January 31, 2021

GROUP CURATION: N/A

The University transitioned to a hybrid model for the spring semester, inviting over 3,000 students back to campus the week of Jan. 10 for the first time since March. According to West Philadelphia residents, the local community was not consulted regarding this decision, which occurred amid ongoing COVID-19 cases that have resulted in over 107,000 cases and 2,800 deaths in the city.

Alex Wiles, a lifetime resident of Philadelphia, said that students should not be on Penn’s campus at all, and blames University leadership for endangering its students, staff, as well as the surrounding community.

The speedy posting of a federal plan bodes well, says Janet Hamilton, the executive director of the Council of State and Territorial Epidemiologists, based in Atlanta, Georgia. “We are really pleased to see the release of a national strategy to ensure we have a coordinated path forward,” she says.

Still, scientists who’ve long been working on the US coronavirus response, say that Biden’s strategy needs more detail, particularly on the funding, staffing and procedures for some initiatives, such as the plan for scaling up surveillance of new variants of the coronavirus, SARS-CoV-2.

The ongoing coronavirus disease 2019 (COVID-19) pandemic has produced widespread changes to our day-to-day lives in 2020. Many of us have spent the year working or going to school from home. Other interventions have been implemented to keep us and others safe when we do venture from home, including the wearing of face masks. Additionally, widespread availability of alcohol-based hand sanitizer, often in automatic, hands-free dispensers, has become commonplace in public places.

In this issue of JAMA Ophthalmology, authors from 2 different countries report eye injuries in children from inadvertent contact with alcohol-based hand sanitizer. Martin et al found a 7-fold increase in ocular exposure in children in 2020, with a corresponding increase in surgeries required to manage severe injuries resulting from these exposures.

Chemical injuries vary in severity, from mild and self-limited to severe and vision-threatening or even globe-threatening. The specter of amblyopia looms over even mild injuries in very young children. Recent work suggests that posterior segment inflammation can accompany ocular surface chemical injuries, further compounding the threat to visual acuity.

The CDC ordered software that was meant to manage the vaccine rollout. Instead, it has been plagued by problems and abandoned by most states.

Prior to yesterday’s release, the [State Profile Reports] had been largely hidden from the public, being released only to state governors by the Trump administration’s coronavirus task force.

“The purpose of this report is to develop a shared understanding of the current status of the pandemic at the national, state, regional and local levels,” the reports state.

The data included in the SPRs notes the numbers of COVID-19 cases, percentages of hospitals with staffing and supply shortages, and the most at-risk counties and localities organized by

President-elect Joe Biden has chosen a research policy maven—and familiar face—to be both his science adviser and head of the White House Office of Science and Technology Policy (OSTP).

Eric Lander, 63, is president and founding director of the Broad Institute, which is jointly run by Harvard University and the Massachusetts Institute of Technology. A mathematician turned molecular biologist, Lander was also co-chair of the President’s Council of Advisors on Science and Technology (PCAST) for 8 years under former President Barack Obama, where he worked closely with Obama’s science adviser, John Holdren, and interacted with Biden.

“Eric is a fabulous choice, and he will make a terrific science adviser,” predicts Holdren, who calls Lander “a science polymath” for his breadth of knowledge across many disciplines. That’s also true for policy, Holdren says. “Eric’s fingerprints were on every one of PCAST’s 39 reports” issued under Obama, Holdren adds, noting that six of them covered previous pandemics and public health crises.

Portugal will focus on adopting this first legal framework at EU level for artificial intelligence, which should be based on a “transparent framework, taking into account the risks involved and protecting the EU’s values, on issues such as human rights and privacy, among others”, he said.

“It is important for citizens to have confidence in the digital economy”, he noted, pointing out that “artificial intelligence has been used in applications for individuals and consumers and in the next steps should focus on processing large volumes of data regarding industry and cities”.

Colby College today announced that it has received a $30-million gift to establish the first cross-disciplinary institute for artificial intelligence (AI) at a liberal arts college. Made possible by the tremendous generosity of the Davis family and trustee of its charitable foundation Andrew Davis ’85, LL.D. ’15, the Davis Institute for Artificial Intelligence will provide new pathways for talented students and faculty to research, create, and apply AI and machine learning (ML) across disciplines while setting a precedent for how liberal arts colleges can shape the future of AI. … The Davis Institute for Artificial Intelligence, which will open this fall following a national search for a founding director, will prepare Colby students for a future where AI is transforming a growing array of industries, careers, and modes of discovery, creativity, and scholarship.

The [Global AI Action Alliance] is a new, multistakeholder collaboration platform designed to accelerate the development and adoption of such tools globally and in industry sectors. The Alliance brings together over 100 leading companies, governments, international organizations, non-profits and academics united in their commitment to maximizing AI’s societal benefits while minimizing its risks.

“AI holds the potential to deliver enormous benefits to society, but only if it is used responsibly. We are launching the Global AI Action Alliance along with our partners to shape a positive, human-centered future for AI at this decisive moment in its development,” said Klaus Schwab, Founder and Executive Chairman of the World Economic Forum

Members of the Alliance work together to identify and implement the most promising tools and practices for ensuring that AI systems are ethical and serve all society members, including groups historically under-represented in the AI ecosystem. Supported by a grant from the Patrick J. McGovern Foundation, the Alliance provides a community for real-time learning and rapid scaling of proven approaches to ethical AI, as well as a forum to accelerate collective action on emerging challenges and issues.

We mere mortals haven’t truly been competitive against artificial intelligence in chess in a long time. It’s been 15 years since a human has conquered a computer in a chess tournament. However, a team of researchers have developed an AI chess engine that doesn’t set out to crush us puny humans — it tries to play like us.

The Maia engine doesn’t necessarily play the best available move. Instead, it tries to replicate what a human would do. The AI emerged as a result of a paper co-authored by researchers from Cornell University, the University of Toronto and Microsoft.

It’s easier to talk about people, companies, or events than to talk about ideas. But one idea worth discussing, despite its complexity, is how artificial intelligence could reorder hotel distribution.

Some researchers are wondering if artificial intelligence could handle some of the more complex tasks of shopping and haggling. Could new algorithms and processes shrink the role of travel search engines and comparison apps? Could the cost of bringing buyers and sellers together shrink thanks to technical innovations?

“What is cumbersome and inefficient is the way we, as consumers, interact with travel businesses,” said Humayun Sheikh, co-founder and CEO of Fetch.ai, a startup based in Cambridge, UK. “We need a shift of the technical architecture in the middle.”

“People didn’t think that that UV would have any impact on the COVID-19 virus, and if it did, it would be very small and minimal,” Mullen said. “We learned how to operate in different ways and we got to the point where we were inactivating in fractions of a second, in thousands of a second.”

The team was able to pull a structure from a product already in development as a starting point for Nanowave Air. The electronics are mechanically contorted in the device—an artifact of their flexible electronics research—to intensify the light. Though UV light can be dangerous, causing sunburns and eye damage, it is completely contained within the device. Nanowave Air can kill 99 percent of the virus in less than two thousandths of a second, and it can inactivate an entire room in only 75 minutes. Mullen said it can blast the inactivated air 11 feet across the room. This helps air circulate around the room, so infected air can better reach the Nanowave Air and become safe again.

The University of San Diego (USD) is proud to announce and welcome Maritza Johnson, PhD, as the director of a new center focused on data science, artificial intelligence and society. In this pioneering role, Johnson will be responsible for developing the center.

Johnson is a widely respected data privacy and security professional who joins USD from Google, where she was most recently a senior user experience researcher. She holds a PhD in computer science from Columbia University and earned her bachelor’s in computer science from USD in 2005. She has lectured at University of California, Berkeley and the University of San Diego, and has published in the areas of data privacy and security.

The University of Illinois Urbana-Champaign announced that it will discontinue its use of remote-proctoring software Proctorio after its summer 2021 term. The decision follows almost a year of outcry over the service, both on UIUC’s campus and around the US, citing concerns with privacy, discrimination, and accessibility.

Proctorio is one of the most prominent software platforms that colleges and universities use to watch for cheating on remote tests. It uses what its website describes as “machine learning and advanced facial detection technologies” to record students through their webcams while they work on their exams and monitor the position of their heads. The software flags “suspicious signs” to professors, who can review its recordings. The platform also enables professors to track the websites students visit during their exams, and bar them from functions like copy / pasting and printing.

USC and Amazon today announced they are creating a joint research center focused on development of new approaches to machine learning (ML) privacy, security, and trustworthiness. The Center for Secure and Trusted Machine Learning, which will be housed at the USC Viterbi School of Engineering, will support USC and Amazon researchers in the development of novel approaches to privacy-preserving ML solutions.

The expectation is that the center will unleash a new line of fundamental research on privacy and security aspects of machine learning – a timely and critical effort given the proliferation of artificial intelligence across all aspects of society from education to finance, transportation, healthcare and many others.

Each year, the center will provide support for research projects focused on the development of new methodologies for secure and privacy-preserving machine learning solutions that can scale to support billions of users. In addition, the center will provide annual fellowships to talented doctoral students working in this research area, enabling them to advance research frontiers. Fellowship recipients will be named as Amazon ML Fellows in recognition of their promise and achievements. These fellowships will give students greater understanding of industry and solution-driven research.

Graph analytics is gaining favor for business applications in which insights must be derived from massive unstructured, connected datasets. But there are different computational issues compared to running analytics on traditional, structured data. RTInsights recently sat down with co-founders Keshav Pingali, CEO, and Chris Rossbach, CTO of Katana Graph. We discussed what makes graph analytics different, how to speed the time to results, and how the company’s partnership with Intel fits into the equation. Here is a summary of our conversation.

Distribution shifts can cause significant degradation in a broad range of machine learning (ML) systems deployed in the wild. However, many widely-used datasets in the ML community today were not designed for evaluating distribution shifts. These datasets typically have training and test sets drawn from the same distribution, and prior work on retrofitting them with distribution shifts has generally relied on artificial shifts that need not represent the kinds of shifts encountered in the wild. In this paper, we present WILDS, a benchmark of in-the-wild distribution shifts spanning diverse data modalities and applications, from tumor identification to wildlife monitoring to poverty mapping. WILDS builds on top of recent data collection efforts by domain experts in these applications and provides a unified collection of datasets with evaluation metrics and train/test splits that are representative of real-world distribution shifts. These datasets reflect distribution shifts arising from training and testing on different hospitals, cameras, countries, time periods, demographics, molecular scaffolds, etc., all of which cause substantial performance drops in our baseline models. Finally, we survey other applications that would be promising additions to the benchmark but for which we did not manage to find appropriate datasets; we discuss their associated challenges and detail datasets and shifts where we did not see an appreciable performance drop. By unifying datasets from a variety of application areas and making them accessible to the ML community, we hope to encourage the development of general-purpose methods that are anchored to real-world distribution shifts and that work well across different applications and problem settings. Data loaders, default models, and leaderboards are available at this https URL.