NYU Data Science newsletter – October 2, 2015

GROUP CURATION: N/A

In a policy statement issued today, the American Statistical Association (ASA) stated statistics is “foundational to data science”—along with database management and distributed and parallel systems—and its use in this emerging field empowers researchers to extract knowledge and obtain better results from Big Data and other analytics projects.

The statement also encourages “maximum and multifaceted collaboration” between statisticians and data scientists to maximize the full potential of Big Data and data science.

… From 2000 to 2014, master’s and doctorate degrees in statistics grew significantly at 260% and 132%, respectively.

Yet demand for statisticians is likely to continue to outpace this growth.

It’s time once again time to officially fire up our favorite piece of sports-related software: the NYT 4th Down Bot. It’s a better bot. We’ve made several updates and improvements, including the ability to announce our analysis before a fourth down, so that you can see it and an N.F.L. coach can act on it.

NCES reveals that the growing data science field can pull from an incredible number of statistics majors, but it’s still not enough.

Increasing by more than 300 percent since the 1990s, the number of undergraduates in statistics degree programs is hurtling the major into the fastest growing in the U.S. But the growth may not be enough to satisfy the high demand for technology-based fields like data science and database management—which could spell trouble for colleges and universities anxious to effectively analyze massive amounts of big data.

After two amazing years with the Nordstrom Data Lab, I’ve accepted a research scientist position at Amazon Web Services to work on S3. I’m excited to begin a new chapter of my career, and relieved that the interview process is over because it’s grueling and time-consuming. Interviews typically consist of one to three screener conversations and then an all-day on-site, and they’re stressful because it’s hard to know what you’ll be asked and often you’re expected to perform feats of intellect that you don’t typically do as a data scientist (at least not devoid of context, from memory and over the phone).

Earlier this week I gave a seminar at UCSF. In addition to my usual scientific spiel, I decided to end my talk with a proposal to UCSF faculty for action that could take make scholarly communication better. This is something I used to do a lot, but have mostly stopped doing since my entreaties rarely produce tangible actions. But I thought this time might be different. I was optimistic that recent attention given by prominent UCSF professor Ron Vale to the pervasive negative effects of our current publishing system might have made my UCSF faculty colleagues open to actually doing something to fix these problems.

So I decided to issue a kind of challenge to them to not just take steps on their own, but to agree collectively to take them together.

This is part I of a multi-part series detailing some of the techniques we’ve used here at Baidu’s Silicon Valley AI Lab to accelerate the training of recurrent neural networks. This part focuses on GEMM performance. Later entries might focus on how we parallelize across GPUs, working with half precision, efficient GPU CTC calculation, GRU and LSTM implementation tricks…

This course is intended to provide an account of proximal tools in convex optimization with a view towards their applications in certain areas of data science (signal and image processing, inverse problems, statistical data analysis, machine learning, etc.). The basic theory will be provided and a strong emphasis will be placed on algorithm design and concrete applications.

Patrick Combettes of Université Pierre et Marie Curie is the instructor. Class sessions are Tues/Thurs 1:25-3:15, from October 27 to December 15, 2015, in WWH 1302.

Eero Simoncelli, a professor in New York University’s Center for Neural Science, has won an Engineering Emmy® Award for the creation of a now widely used algorithm that assesses how viewers perceive the quality of an image or video, the Television Academy announced this week.