NYU Data Science newsletter – February 2, 2016

GROUP CURATION: N/A

I have been in the profession for about six years now. During this time, subject specialists in academic libraries are mostly trying to articulate our new roles in data and information management, compared to the traditional roles of information gatekeepers. In-depth engagement in the research and learning experiences of our user community truly sets our current role apart from the traditional role of the librarian sitting behind the circulation desk handing out books. We are now striving to be facilitators of academic success and interdisciplinary collaborations.

Taking a page from Jules Verne, researchers at Microsoft believe the future of data centers may be under the sea.

Microsoft has tested a prototype of a self-contained data center that can operate hundreds of feet below the surface of the ocean, eliminating one of the technology industry’s most expensive problems: the air-conditioning bill.

So I’ve arrived late at the office of Viv, an artificial intelligence company based in San Jose, California. I missed my train from San Francisco after dawdling leaving my apartment and then finding the taxi service app on my phone wouldn’t work. Dag Kittlaus, who I’ve kept waiting, looks on the bright side. “Your trials of getting here are a perfect illustration of how Viv will be helpful,” he says. “Wouldn’t it be nice to say ‘I need to get to San Jose, give me my options’ and Viv would know how close you are to the train station, when the next train is coming, where the nearest cars, how much it was going to cost…”

Kittlaus is the co-founder and CEO of Viv, a three-year-old AI startup backed by $30m, including funds from Iconiq Capital, which helps manage the fortunes of Mark Zuckerberg and other wealthy tech executives. In a blocky office building in San Jose’s downtown, the company is working on what Kittlaus describes as a “global brain” – a new form of voice-controlled virtual personal assistant. With the odd flashes of personality, Viv will be able to perform thousands of tasks, and it won’t just be stuck in a phone but integrated into everything from fridges to cars. “Tell Viv what you want and it will orchestrate this massive network of services that will take care of it,” he says.

Yoshua Bengio leads one of the world’s preëminent research groups developing a powerful AI technique known as deep learning. The startling capabilities that deep learning has given computers in recent years, from human-level voice recognition and image classification to basic conversational skills, have prompted warnings about the progress AI is making toward matching, or perhaps surpassing, human intelligence. Prominent figures such as Stephen Hawking and Elon Musk have even cautioned that artificial intelligence could pose an existential threat to humanity. Musk and others are investing millions of dollars in researching the potential dangers of AI, as well as possible solutions. But the direst statements sound overblown to many of the people who are actually developing the technology. Bengio, a professor of computer science at the University of Montreal, put things in perspective in an interview with MIT Technology Review’s senior editor for AI and robotics, Will Knight.

IF social scientists and policy makers have learned anything about how to help the world’s poorest people, it’s not to trust our intuitions or anecdotal evidence about what kinds of antipoverty programs are effective. Rigorous randomized evaluations of policies, however, can show us what works and what doesn’t.

Consider microloans. This celebrated and sensible-sounding strategy — advancing small sums of money to help women in developing countries become entrepreneurs — has been supported by inspiring tales of people pulling themselves out of poverty by creating small businesses. But studies show that such stories are outliers. Six randomized evaluations of microloan programs, for example, published last year in the American Economic Journal, found that microloans, though helpful for the poor, didn’t actually increase income for the average borrower.

Two other recent studies, conducted in sub-Saharan Africa by field researchers working with scholars of behavioral science in the United States and England, also tested antipoverty strategies and found in each case that conventional instincts about what would work were wrong.

Imagine a list containing every single Wawa location on the planet. Such an enormous amount of information seems impossible to conceive — but soon, computer science students at Penn will begin to make some sense of it.

This semester, the Computer and Information Sciences Department introduced a new course called “Special Topics: Foundations of Data Science.” The course is seminar-style, meeting only once a week on Thursday for a three-hour block. Prerequisites for interested students are Math 240 or above and some significant exposure to probability.

After Pooja Ramesh left her engineering job at Intel Corp. to get married and move to Denver two years ago, she took her time finding a new gig.

In fact, she spent an extra 12 weeks last fall training for a career that has become one of the hottest jobs of 2016: data scientist.

“Everything clicked,” said Ramesh, who used part of her time at the Galvanize data-science program to explore how data science can speed up the detection of breast cancer. “I was doing statistical analysis, but I didn’t know it fell under data science. In coming to the open house (last fall), I realized I’d been doing this.”

BIDS Data Science Lecture Series … Speaker: Peter Aldhous, Science Reporter, BuzzFeed

I’ll talk about the history of data journalism; its current practice; and its challenges, including “dirty” data, unstructured text, forensic image analysis, and the application of machine learning. [59:17]

When it comes to data, Stephen Parente’s motto is simple: bigger is better. The economist reached this position over the course of a long career in health information technology—both in and outside of academia. Recently, his health-economics consultancy was tapped to advise the Centers for Medicare and Medicaid Services, in charge of implementing Medicare fraud-mitigation practices.

“Double down on Big Data,” Parente says. “I understand that people have issues with privacy, but if they understood that having this data be shared—which most people are OK with; their data is being shared when they shop online—then they will be hassled less and the system will be made whole.”

… Unfortunately, the focus on the low-end approach of disruption is actually clouding our ability to spot the things that are: cheaper, more accessible, and built on an advantaged cost structure. Specifically, it appears that data-enabled disruptors often confound industry pundits. To get a sense for the point, just look to a few highly contested examples.

Is Uber disruptive? The wrong answer would be to say, “No, because their first product started in the high end of the market.” The right answer would be to acknowledge that the platform they ultimately launched allowed them to add lower cost drivers (in the form of UberX) and offer cheaper, more accessible, transportation options with a structural cost advantage to both taxi services and potentially even car ownership. The convenience of the app is only the most obvious, and easiest to copy, factor.

Scientific Data is inviting submissions releasing and describing datasets related to Zika virus and the associated outbreak of microcephalic cases in South America. Submissions may be considered for inclusion in a special article collection on this topic.

To encourage early release of data related to this serious public health issue, all submissions received by 31st May 2016 and ultimately accepted for publication will have their article processing charge waived in full. Submissions after this date are still welcome, and will be considered according to the standard policies of the journal.

Wordbank is an open database of information about children’s vocabulary growth.

Wordbank archives data from the MacArthur-Bates Communicative Development Inventory (MB-CDI), a family of parent-report questionnaires. Wordbank enables researchers to analyze MB-CDI data in terms of aggregate vocabulary, individual items, demographic variables, and more. It provides interactive visualizations, exploratory reports, and data export tools.