Data Science newsletter – November 4, 2019

GROUP CURATION: N/A

JPMorgan Chase & Co.’s new Manhattan headquarters is meant to be an ode to both the company and the city — a monumental glass-and-steel tower that says the nation’s largest bank grew up here. But New York may be losing its luster.

Despite more than two centuries of history in a city synonymous with the global financial industry, JPMorgan is quietly shrinking its workforce there. The bank’s been building up its presence in other locations and is now considering relocating several thousand New York-based employees out of the area to help rein in costs ahead of a possible economic downturn, according to people with knowledge of the bank’s strategy.

By some time next year, predicts Raghua Ramakrishnan, there will be more than 20 billion connected electronic devices in the world.

“The amount of data exchanged by sensors will be more than the amount exchanged by human beings,” said the CTO for Data and a Technical Fellow at Microsoft, speaking at the third annual Rice Data Science Conference. The Oct. 14-15 event was hosted by the Ken Kennedy Institute for Information Technology and drew some 340 leaders from industry and academia.

“Data science is a field in which scientists work with domain experts to address complex problems beyond what was thought possible even five years ago,” said Lydia Kavraki, director of the Ken Kennedy Institute and Noah Harding Professor of Computer Science at Rice.

While searching for the person who grifted me in Chicago, I discovered just how easy it is for users of the short-term rental platform to get exploited.

Sudip Parikh has helped shape U.S. science policy as a staffer on a powerful congressional spending panel. He’s been a senior health care executive for a large nonprofit organization that manages several federal research facilities. And in January 2020, the 46-year-old structural biologist will become the new CEO of AAAS (which publishes Science) as the 171-year-old association pursues its mission to advance science and serve society.

“It’s a marvelous organization, and I’m super excited to become a part of it,” Parikh says. “I think every scientist has a place in their heart for AAAS. My goal is to turn those warming feelings into a valuable engagement with AAAS that will help us move forward.”

In January, a team of researchers from the London School of Economics released a report that examined children’s data and online privacy. Their study found that surveillance had several negative side effects for children. “It might obstruct children’s development of important skills,” the researchers wrote, “and it could affect negatively the trust relationship between the parent and the child.”

In a new paper, a team of USC computer scientists and psychologists teamed up to investigate how music affects how you act, feel and think

Two years ago, volunteer insectologists sounded the alarm — the number of flying insects had drastically fallen. Now a new study on three protected regions in Germany confirms these fears.

In a smart move, the social media company has enlisted the help of health experts to make this happen. Facebook teamed up with four national medical groups to launch this initiative: the American Heart Association, American Cancer Society, American College of Cardiology, and the Centers for Disease Control and Prevention. The feature’s recommendations are powered by content that has been scientifically researched and proven and are provided in an easy to understand format without medical jargon.

“The goal here is to take the amazing wisdom of our partner organizations and deliver it in a consumer-friendly way to the pockets of our entire Facebook community so they can engage in its preventive health recommendations and advice and become their own health advocates,” said Dr. Freddy Abnousi, Facebook’s head of healthcare research.

At NCSA, researchers are working to apply machine learning across multiple disciplines, making research more efficient than ever before. … The promise of fast and integrated detection is to find currently-undetected events hidden in existing data streams, and to better understand them by using multiple instruments together. To address these challenges, NCSA is participating in three new National Science Foundation (NSF) awards, two of which are led by NCSA teams, to advance multi-messenger astrophysics as a whole.

Africa has long been regarded as the cradle of humankind, but scientists seeking a more specific location have narrowed in on northern Botswana as the “homeland” for all modern humans, according to a new study.

There, south of the Greater Zambezi River Basin, which includes northern Botswana and parts of Namibia and Zimbabwe, the ancestors of Homo sapiens began 200,000 years ago, the researchers said.

Their new study, published Monday in the journal Nature, suggests that the ancestors of modern humans thrived for 70,000 years in this region before climate change led them to migrate out of Africa and eventually span the globe.

Fires, floods, polar vortexes and hurricanes — every season brings another disaster seemingly linked to climate change. But natural disasters happened before climate change, too. So how are we supposed to know which disasters are fated because of the stars, and which are fated because of 100 years of global CO2 emissions?

Since the calamities aren’t going to stop anytime soon, we figured it’d be helpful to give you a guide for how to think empirically about disaster news in a climate change era. We asked some climate scientists what they think the headlines get wrong or leave out. They offered four tips for thinking about natural disasters and climate change — tips they say can make the difference between feeling hopeless about the future and finding ways to change it.

1) No disaster is actually “natural”

The Sam M. Walton College of Business at the University of Arkansas is now offering a Master of Science in Economic Analytics, a 10-month program. Classes will begin in fall 2020 and applications are being accepted.

CDL was founded in 2012 by Ajay Agrawal at the Rotman School of Management at the University of Toronto. The accelerator operates out of five universities across Canada and in Europe, at the University of Oxford’s Saïd Business School. CDL does not require equity or fees from participating companies and receives most of its funding through private donors. Recently, the program hosted its Machine Learning and the Market for Intelligence conference in Toronto, which focused on how AI will affect power and society.

“The types of founders that join our program are often engineers or scientists. They have deep expertise in their technical domain, but lack business judgment,” said Agrawal. “They are regularly faced with a to-do list of hundreds of things they could be doing to build their business, but don’t have the cycles to do all those things, and so they have to pick from the list. Judgment is the skill of prioritizing the list.”

Students are campaigning for Cambridge University to sever its links with Palantir, a technology company that supplies software for controversial US government immigration enforcement programmes.

On Monday, students from the #NoTechForTyrants campaign protested a talk, entitled ‘Privacy and Civil Liberties (PCL) & Machine Learning’, held by senior Palantir employees at the Department of Computer Science and Technology. Palantir directs money towards the Department through its sponsorship of student prizes and as a member of the Supporters’ Club, for which it pays an annual fee.

One third of all approved drugs target GPCRs in one way or the other. However, for about one third of GPCRs the scientific community does not know which molecules – or ligands as they are called – bind to the receptor and activate a signal.

Now researchers from the University of Copenhagen, along with collaborators from University of North Carolina at Chapel Hill and Thomas Jefferson University have discovered new signaling systems: five GPCRs activated by 17 peptide ligands. The new study, which is published in Cell, therefore expands the known peptide-GPCR network from 348 to 415 interactions or by 19 percent in total.

San Francisco, CA November 8-9. “A few months back, as we began planning the November issue, we started to feel that national malaise, the distress that surrounded the environment, health, cybersecurity, politics. But we also knew that there were really smart people out there, working to craft a better future. We decided to write about them and invite them (and many more like them) to come talk to us.” [$$$]

Currently used metrics for assessing summarization algorithms do not account for whether summaries are factually consistent with source documents. We propose a weakly-supervised, model-based approach for verifying factual consistency and identifying conflicts between source documents and a generated summary. Training data is generated by applying a series of rule-based transformations to the sentences of source documents. The factual consistency model is then trained jointly for three tasks: 1) identify whether sentences remain factually consistent after transformation, 2) extract a span in the source documents to support the consistency prediction, 3) extract a span in the summary sentence that is inconsistent if one exists. Transferring this model to summaries generated by several state-of-the art models reveals that this highly scalable approach substantially outperforms previous models, including those trained with strong supervision using standard datasets for natural language inference and fact checking. Additionally, human evaluation shows that the auxiliary span extraction tasks provide useful assistance in the process of verifying factual consistency.

Spleeter is the Deezer source separation library with pretrained models written in Python and uses Tensorflow. It makes it easy to train source separation model (assuming you have a dataset of isolated sources), and provides already trained state of the art model for performing various flavour of separation.”

“In the case of audio fingerprinting, the fingerprinting is based on the device’s audio stack. Just as canvas fingerprinting takes advantage of the Canvas API, the technology that makes audio fingerprinting possible is an API called the AudioContext API. It is an interface of the Web Audio API that is a part of most modern browsers.”