Data Science newsletter – May 7, 2020

GROUP CURATION: N/A

The blurring of work and home lives might not be just a temporary side effect of the pandemic. Though children will (hopefully) go back to school in the fall, many white-collar workers will be strongly encouraged to work from home at least some of the time over the next year or so. While working remotely confers some mental-health and other benefits, the “job” as we know it might never be the same. Conferences, in-person meetings, and even handshakes might be deemed not worth the risk of infection. What might emerge is a future in which results-oriented introverts prevail while those who thrive on face-to-face interactions and office politics fumble. In the post-pandemic workplace, nerds may get their revenge.

We’re releasing an analysis showing that since 2012 the amount of compute needed to train a neural net to the same performance on ImageNet1 classification has been decreasing by a factor of 2 every 16 months.

UC San Francisco is launching a workforce training and technical assistance program on Thursday, May 6, 2020, in partnership with the California Department of Public Health, to facilitate the training of thousands of individuals across the state in public health techniques and strategies, including contact tracing, case investigation and administration, to limit the ongoing spread of COVID-19. The move represents the next stage of the state’s effort to bring the COVID-19 epidemic under control and prepare the communities for a safe emergence from shelter in place.

The pandemic of coronavirus disease 2019 (COVID-19) continues to affect much of the world. Knowledge of diagnostic tests for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is still evolving, and a clear understanding of the nature of the tests and interpretation of their findings is important. This Viewpoint describes how to interpret 2 types of diagnostic tests commonly in use for SARS-CoV-2 infections—reverse transcriptase–polymerase chain reaction (RT-PCR) and IgM and IgG enzyme-linked immunosorbent assay (ELISA)—and how the results may vary over time (Figure).

The effort is looking to enroll 200,000 to 250,000 device owners, and will support clinical evaluation and regulatory submissions of the company’s PPG atrial fibrillation algorithm.

Australians running the Covidsafe contact tracing app on iPhones may not be recording all the data required if they don’t have the app running in the foreground or they are using an older model phone, the government has admitted.

More than 5.1 million Australians have downloaded and registered to use the app on iPhone and Android devices, and, while the Android version works while running in the background (ie, not open on the screen), the iPhone version works best when the app is open on the screen and the phone is unlocked.

After nearly a year of uncertainty, the department has drafted a new rule to address the issue, two sources told Reuters. The rule, which could still change, essentially allows U.S. companies to participate in standards bodies where Huawei is also a member, the sources said.

The draft is under final review at the Commerce Department and, if cleared, would go to other agencies for approval, the people said. It is unclear how long the full process will take or if another agency will object.

“As we approach the year mark, it is very much past time that this be addressed and clarified,” said Naomi Wilson, senior director of policy for Asia at the Information Technology Industry Council (ITI), which represents companies including Amazon.co Inc, Qualcomm Inc and Intel Corp.

Chris Atkeson, a professor at the Robotics Institute at Carnegie Mellon University, watches more scenes featuring robots from movies and television and continues to break down how accurate their depictions really are. Does Hal from “2001: A Space Odyssey” hate humans? Would robots speak to each other like R2-D2 and C-3PO in “Star Wars”? [video, 20:13]

Users diagnosed with the virus are asked to upload their close contact information to what the government describes as “a highly secure information storage system” located on Amazon Web Services. This enables health officials to look up people who have been diagnosed, find all the COVIDSafe users they have recently come into contact with, and advise them what to do.

Suranga Seneviratne, a lecturer on computer security at the University of Sydney, believes the app poses no major security risks as it cannot access sensitive data.

Google told employees they cannot expense free food or repurpose unused budgets from things such as cancelled events on perks such as fitness.

The company added that workers can’t use extra funds to donate to charities or fundraisers of their choice.

“Hello Monday” host and LinkedIn Senior Editor at Large Jessi Hempel takes over as host for a day, interviewing Patrick [McGinniss] about how he has explored the world to observe how FOMO and FOBO arise as a function of culture, social status, and even as our response to crisis, such as during the current Covid-19 pandemic. [audio, 28:28]

The long-simmering debate over facial recognition technology is taking on new urgency during the pandemic, as companies rush to pitch face-scanning systems to track the movements of Covid-19 patients.

That’s playing out in California, where state legislators on Tuesday will debate legislation that would regulate the use of the technology. Its most controversial element: It would permit companies and public agencies to feed people’s facial data into a recognition system without their consent if there is probable cause to believe they’ve engaged in criminal activity. The bill isn’t specifically meant for the coronavirus response, but if enacted, could shape the way that people with Covid-19 and their contacts are tracked and traced in the coming months.

The legislation has won the support of Microsoft, but it has garnered opposition from more than 40 civil rights and privacy groups and from 18 public health scholars.

Facebook revealed the members of its first ever oversight board. The board will eventually swell to nearly double its current size over time, Facebook says. Here are the inaugural 20 members:

1. Tawakkol Karman

Baltimore, Boston, and Washington, D.C. all recently joined the growing list of cities imposing mandatory caps on the commission fees third-party delivery services charge restaurants. San Francisco, Chicago, NYC, and Los Angeles have already passed similar measures or are considering them.

The D.C. Council passed emergency COVID-19 legislation on Tuesday that, among other things, capped commission fees at 15 percent during the city’s state of emergency. As the Washington Post noted, “The commission cap, similar to ones implemented in Seattle and San Francisco, is meant to help eateries turn profits on those sales.”

U.S. primary insurance giant Allstate has added roughly $221 million of new protection to its core nationwide excess of loss catastrophe reinsurance program in the first-quarter of 2020, with the capital markets playing an important role through the Sanders Re II catastrophe bond.

Allstate has pushed up the top of its nationwide excess catastrophe reinsurance tower to almost $5 billion with the changes, adding $121 million to this tower and keeping its per-occurrence retention at an unchanged $500 million.

Online May 13, starting at 10 a.m. PDT. “As plans for re-opening businesses, communities, and schools emerge, mechanisms to track the SARS-COV-2 virus become increasingly critical to consider. In this conversation led by Nobel Laureate Saul Perlmutter, Director of the Berkeley Institute for Data Science and Professor of Physics, Berkeley faculty will present their recent research findings and data on COVID-19 infection and death rates.” [live web broadcast]

“Post your studies on http://ChildrenHelpingScience.com!”

“The secret sauce here is BASNet (Qin et al, CVPR 2019) for salient object detection and background removal.”

“The accuracy and range of this model are stunning and there are many nice use cases so I packaged it as a micro-service / docker image.”

At DataKind, we have seen how relatively simple techniques can empower an organization. For example, working with Conservation International, volunteer data scientists from DataKind launched Colandr, a computer-assisted evidence synthesis tool that helps surface evidence relevant to a user-identified topical area, and helps researchers, practitioners, and policy makers find resources to make evidence-based decisions. To date, Colandr has already been adopted by over 200 other organizations, a good example—like many others—that showcases how mission-driven organizations can collaboratively design innovative and impactful solutions to address tough social challenges.

Here we present six mature, accessible NLP techniques, along with potential use cases and limitations, and access to online demos of each (including project data and sample code for those with a technical background).

“A unique collaboration between WHO, the London School of Hygiene and Tropical Medicine, ACAPS, University of Oxford, Global Public Health Intelligence Network, U.S. Centers for Disease Control and Prevention and the Complexity Science Hub Vienna has brought these datasets together, using a common taxonomy and structure, into a single, open-content dataset for public use.”

Topic models are a useful analysis tool to uncover the underlying themes within document collections. Probabilistic models which assume a generative story have been the dominant approach for topic modeling. We propose an alternative approach based on clustering readily available pre-trained word embeddings while incorporating document information for weighted clustering and reranking top words. We provide benchmarks for the combination of different word embeddings and clustering algorithms, and analyse their performance under dimensionality reduction with PCA. The best performing combination for our approach is comparable to classical models, and complexity analysis indicate that this is a practical alternative to traditional topic modeling.