The mission of the social sciences is to understand and ameliorate society’s greatest challenges. The data held by private companies, collected for different purposes, hold vast potential to further this mission. Yet, because of consumer privacy, trade secrets, proprietary content, and political sensitivities, these datasets are often inaccessible to scholars. We propose a novel organizational model to address these problems. We also report on the first partnership under this model, to study the incendiary issues surrounding the impact of social media on elections and democracy: Facebook provides (privacy-preserving) data access; eight ideologically and substantively diverse charitable foundations provide funding; an organization of academics we created, Social Science One (see SocialScience.One), leads the project; and the Institute for Quantitative Social Science at Harvard and the Social Science Research Council provide logistical help.
University of Chicago, Department of Computer Science
from
Over three-quarters of today’s internet traffic comes from streaming video, a number that is only projected to rise over time. To meet this demand, internet service providers offer consumers faster data speeds at premium prices, with gigabit-per-second tiers available in some areas. But do these pricier plans actually improve the quality of video streaming? A Wall Street Journal investigation, published August 20th, answered this question with the help of UChicago Professor Nick Feamster, in a collaboration that both informs consumers and advances science.
The story, “The Truth About Faster Internet: It’s Not Worth It,” began three years ago with what the reporters thought would be a simple question: do faster speeds matter for streaming video? To answer it, they turned to Feamster, then at Princeton University, for his expertise on networked computer systems. Feamster’s group had developed software systems that could reliably measure the actual internet speeds consumers received at home, one important part of the WSJ inquiry. But gathering the other half of the data — the performance of streaming video services in those homes — would require a mix of systems, machine learning, and recruitment.
Nvidia today announced that it has been working with VMware to bring its virtual GPU technology (vGPU) to VMware’s vSphere and VMware Cloud on AWS. The company’s core vGPU technology isn’t new, but it now supports server virtualization to enable enterprises to run their hardware-accelerated AI and data science workloads in environments like VMware’s vSphere, using its new vComputeServer technology.
20 years after acquiring Thinking Machines, AI and ML have remained among Oracle’s best-kept secrets. Oracle has been internally working with ML to enhance its products, but now it is starting to come out of its shell with the type of product portfolio that one would expect from an enterprise technology provider.
Sage Bionetworks hosted its 2019 Assembly in Seattle to mark its 10th anniversary. The day-long scientific symposium featured speakers who will helped us honor our history by guiding a conversation about the next generation of open science. The Assembly took place on July 25, 2019, at Bell Harbor Conference Center in Seattle. Learn more at http://sageassembly.org. Panelists: Deborah Estrin, Cornell Tech | Cliff Lynch, Coalition for Networked Information | Carly Strasser, Fred Hutchinson Cancer Research Center | Dario Taraborelli, Chan Zuckerberg Initiative. Moderator: Kara Woo, Sage Bionetworks [video, 1:10:21]
Although logging (both legal and illegal) and other activities also drive deforestation in the Amazon, animal agriculture is the leading cause by far. The World Bank reported that cattle ranching occupies 80 percent of all converted lands in the Amazon rainforest.
Many types of information disorder exist online, from fabricated videos to impersonated accounts to memes designed to manipulate genuine content.
Automation and microtargeting tactics have made it easier for agents of disinformation to weaponize regular users of the social web to spread harmful messages.
Much research is needed to understand the effects of disinformation and build safeguards against it.
In 2018, Facebook announced a partnership to provide data to academics to “help people better understand the broader impact of social media on democracy — as well as improve our work to protect the integrity of elections.”
In April of this year, the first batch of winning proposals was announced. “The urgency of this research cannot be overstated,” wrote the founders of Social Science One, the entity that operates the program. It describes itself as “an LLC operating on a not-for-profit basis.”
But as of today, many of the academic teams remain on hold because Facebook has yet to provide key data required to conduct research into sharing patterns of fake and polarized news, among other projects. Facebook has also declined to provide some of the data it originally said it would offer, citing privacy concerns.
Until recently, the University of Virginia had two computing services teams for researchers: Advanced Research Computing Services for the academic side, and the School of Medicine Research Computing Group on the medical side. The Office of the Vice President for Information Technology coordinated both.
That office this month merged the two groups into what now is called UVA Research Computing, directed by Rick Downs, who had previously managed the academic entity.
The Daily Californian student newspaper, Kate Finman
from
A team of scientists from UC Berkeley, UC San Diego and the University of Washington won a five-year $1 million dollar grant last month from the National Science Foundation, or NSF, to develop CloudBank, a program that will simplify public cloud access for researchers and educators.
The grant went into effect Aug. 1 and will allow scientists to develop a suite of software, outreach, training materials and aids for negotiating and delivering public cloud services — which will help scientists more easily navigate the cloud and learn how to best use their resources, according to the NSF.
“Every semester thousands of students in dozens of courses have all the computing power they might need by simply opening a notebook in a browser,” said UC Berkeley researcher David Culler in an email. “Berkeley’s role in CloudBank is to make it possible for universities throughout the country to bring this sort of capability to their students.”
Two contrasting approaches to AI chips emerged at Hot Chips Conference this week. While Cerebras is building the largest chip ever, other vendors are looking to see how they can disaggregate the larger dies.
Germany, France and Japan have joined forces to fund research into “human-centered” artificial intelligence that aims to respect privacy and transparency, in the latest sign of a global split with the U.S. and China over the ethics of AI.
The three countries’ funding agencies have put out a joint call for research proposals, backed by an initial 7.4 million euros ($8.2 million). They stressed that they “share the same values” and warned that the technology has the potential to “violate individual privacy and right to informational self-determination.”
Observers see the move as part of a wider divergence in AI research priorities, with Europe, plus Japan and potentially Canada, taking the lead on its ethical development.
For the Department of Agriculture’s Economic Research Service (ERS), virtually every piece of news reflects a worsening situation. In the latest wrinkle, $25,000 buyout payments, offered to ERS employees in June, were abruptly reduced to $10,000, according to an internal memo obtained by Government Executive.
These “Voluntary Separation Incentive Payments,” or VSIP, are being offered to ERS employees who a) completed a program application and b) were determined by USDA to be eligible. The buyouts are supposed to provide compensation for workers who do not want to relocate to Kansas City this fall.
University of Michigan, Center for Integrative Research in Critical Care
from
Ann Arbor, MI October 18, starting at 8 a.m., University of Michigan. “The 2019 Joyce Massey TBI Summit brings together leading scientists and key opinion leaders to present cutting-edge research, exchange ideas, and discuss the future of [traumatic brain injury] research.” [registration required]
Kansas City, MO September 9, starting at 6 p.m., VeriShip (8880 Ward Pkwy #300). “Data Science KC and the KC Graph Database Meetup join forces to explore the power of connected data. Jennifer Reif, Developer Relations Engineer at Neo4j, will explain how a graph database works and show how you can apply network science to solve real-world problems.” [rsvp required]
Zurich, Switzerland September 2-4. “This is the third in a series of three symposia that discuss societal challenges in computational social sciences. In 2019, the focus will be on “Polarization and Radicalization” (Zurich, 2019).” [$$]
The FTC’s recently created Technology Task Force is looking for an experienced Technology Fellow to assist in our non-public investigations of complex technology-driven products and services. The Technology Fellow will help to analyze and evaluate the technical aspects of current products and services, as well as new ones under development. This would include key business functions enabled by operating systems, software, algorithms, database structures, and API functions, among others. The fellowship would last for one year, with a possible one-year extension.Curious? Read more about the opportunity here, or call Daria Kun at (202) 326-3545 with questions. If you’re ready to apply, send a resume and cover letter to BCRecruit@ftc.gov, subject line “TTF Tech Fellow” by August 30, 2019.
Vancouver, BC, Canada December 13 or 14, at NeurIPS 2019. “Submissions are solicited for the NeurIPS 2019 Joint Workshop on AI for Social Good for unpublished completed work and current work within the social impact space.” Deadline for submissions is September 6.
“The Third Polar Data Forum (PDF III) will be hosted by the Finnish Meteorological Institute at their Dynamicum campus in Helsinki from November 18th to 22nd.” Deadline for abstract submission is September 6.
“Researchers may have just arrived home from a fantastic ACL conference in Florence, but it is already time to start thinking about Seattle 2020. The First Call for Papers is now out! And your paper is due on Monday, December 9! This first in a series of ACL 2020 PC blog posts is meant to inform you about a couple of key changes to look out for in this year’s call.”
This post expands on the NAACL 2019 tutorial on Transfer Learning in NLP.
The tutorial was organized by Matthew Peters, Swabha Swayamdipta, Thomas Wolf, and me. In this post, I highlight key insights and takeaways and provide updates based on recent work.