NYU Data Science newsletter – February 11, 2016

GROUP CURATION: N/A

Last week’s finding by a team of data scientists at Facebook that everyone in the social network is connected by an average of 3.5 “intermediaries” has renewed interest in the longstanding “Six Degrees of Separation” hypothesis: that everyone in the world is connected by some short chain of acquaintances. Not surprisingly, the attention has focused on the plausible assertion that online social networks like Facebook have made the world smaller: that what used to be six degrees is now almost half that. But really what it has revealed is how little we understand this intriguing phenomenon and what it might mean for our world.

… Government databases are filled with everything from traffic data to pet-ownership statistics, and many agencies lack the necessary staff and infrastructure to maintain and analyze all of this information. Public-sector data analysts report that they spend 47 percent of their time collecting and organizing data but less than a third of their time actually gleaning actionable insights from it.

A primary cause of government data hoarding is the public sector’s fragmentation: Data is segregated into specific departmentalized systems and in most cases cannot be compared or analyzed across entire organizations.

UW CSE professor Yejin Choi, an expert in natural language processing, was selected as one of 10 young scientists to watch in the field of artificial intelligence in the latest issue of IEEE Intelligent Systems. The list, which is published biennially, celebrates rising stars in the field and is based on nominations by senior AI researchers in academia and industry.

… The challenge of actually tracing the individual trajectories of group-traveling animals in the wild has kept the available data sparse.

“Technology is about to change this,” says SFI Omidyar Fellow Andrew Berdahl. He has been awarded a National Science Foundation grant to use airborne drones to study a caribou herd as it travels from its summer territory on Victoria Island above the Arctic Circle to its winter grounds on mainland Canada.

Backchannel has assembled a panel of security professionals from technology companies and academia for a weeklong virtual roundtable discussion. This week we’re asking them to look up from their daily battles and fix their eyes on the future. What will it take to make the next decade safer than the last?

On the evening of November 9th, 1989, the Cold War came to a dramatic end with the fall of the Berlin Wall. Four years ago another wall began to crumble, a wall that arguably has as much impact on the world as the wall that divided East and West Germany. The wall in question is the network of paywalls that cuts off tens of thousands of students and researchers around the world, at institutions that can’t afford expensive journal subscriptions, from accessing scientific research.

On September 5th, 2011, Alexandra Elbakyan, a researcher from Kazakhstan, created Sci-Hub, a website that bypasses journal paywalls, illegally providing access to nearly every scientific paper ever published immediately to anyone who wants it. The website works in two stages, firstly by attempting to download a copy from the LibGen database of pirated content, which opened its doors to academic papers in 2012 and now contains over 48 million scientific papers. The ingenious part of the system is that if LibGen does not already have a copy of the paper, Sci-hub bypasses the journal paywall in real time by using access keys donated by academics lucky enough to study at institutions with an adequate range of subscriptions. This allows Sci-Hub to route the user straight to the paper through publishers such as JSTOR, Springer, Sage, and Elsevier. After delivering the paper to the user within seconds, Sci-Hub donates a copy of the paper to LibGen for good measure, where it will be stored forever, accessible by everyone and anyone.

On the show we talk about his creative process to develop ggplot2, its growing popularity, other libraries he has built in the R ecosystem, and strategies for creating popular software for data analysis and visualization. [audio, 1:01:04]

Each year, after the President’s State of the Union address, the Office of Management and Budget releases the Administration’s Budget, offering proposals on key priorities and newly announced initiatives. For the last two years of the Obama administration, we are releasing all of the data included in the President’s Budget in a machine-readable format here on GitHub.

The post was co-authored by Sam Gross from Facebook AI Research and Michael Wilber from CornellTech.

In this blog post we implement Deep Residual Networks (ResNets) and investigate ResNets from a model-selection and optimization perspective. We also discuss multi-GPU optimizations and engineering best-practices in training ResNets. We finally compare ResNets to GoogleNet and VGG networks.

We release training code on GitHub, as well as pre-trained models for download with instructions for fine-tuning on your own datasets.