NYU Data Science newsletter – January 15, 2016

GROUP CURATION: N/A

Looking for tiny facial tics from opponents has been used by poker players for decades as a way to spot a bluff, but now Chinese scientists are using the same technique to develop a lie detector.

They have created an online polygraph that uses facial recognition to study minute changes in facial expression, skin colour, temperature and heart rate – all of which could reveal a lie.

Imagine classrooms outfitted with cameras that run constantly, capturing each child’s every facial expression, fidget, and social interaction, every day, all year long.

Then imagine on the ceilings of those rooms infrared cameras, documenting the objects that every student touches throughout the day, and microphones, recording every word that each person utters.

Picture now the children themselves wearing Fitbit-like devices that track everything from their heart rates to their time between meals. For about a quarter of the day, the students use Chromebooks and learning software that track their every click and keystroke.

What you’re seeing is the future of K-12 education through the eyes of Max Ventilla, the CEO of AltSchool, a Bay Area startup that represents the most aggressive, far-reaching foray into the world of big data and analytics that the K-12 education sector has seen to date.

Looker, a startup selling data analytics software, is announcing today a $48 million funding round led by Kleiner Perkins Caufield & Byers.

Looker can visualize data from many sources, including Amazon Web Services (AWS) Redshift, Cloudera Impala, HP Vertica, MySQL, PostgreSQL, and Snowflake. It’s available as on-premises software or as a cloud service.

It’s that time of the year again—welcome to the Google Tracker! This is a bi-annual series where we recap every ongoing project (that we know about, at least) inside of Google’s sprawling empire.

Though from now on, perhaps we should say, “Every ongoing project inside of Alphabet’s sprawling empire.” “Google” is now a mere company inside of “Alphabet,” the newly formed umbrella company created by Google’s founders. Most of the Google side projects we’ve been tracking in the past have been spun off into companies inside of Alphabet, but rest assured we’re still keeping track of everything.

As always, the Tracker is a big roundup of previous announcements, rumors, and a bit of speculation.

As part of a plan to greatly increase the scale, scope and impact of computer science research and education across the University community, the University of Chicago has appointed prominent data science scholar Michael Franklin to chair its Department of Computer Science and to serve as senior advisor to the provost on computation and data science.

There’s mounting evidence suggesting that student evaluations of teaching are unreliable. But are these evaluations, commonly referred to as SET, so bad that they’re actually better at gauging students’ gender bias and grade expectations than they are at measuring teaching effectiveness? A new paper argues that’s the case, and that evaluations are biased against female instructors in particular in so many ways that adjusting them for that bias is impossible.

It is hard not to be enamoured by deep learning nowadays, watching neural networks show off their endless accumulation of new tricks. There are, as I see it, at least two good reasons to be impressed:

Neural networks can learn to model many natural functions well, from weak priors.

Neural networks can learn surprisingly useful representations

Last month’s International Conference of Computer Vision (ICCV) was full of Deep Learning techniques, but before we declare an all-out ConvNet victory, let’s see how the other “non-learning” geometric side of computer vision is doing. Simultaneous Localization and Mapping, or SLAM, is arguably one of the most important algorithms in Robotics, with pioneering work done by both computer vision and robotics research communities. Today I’ll be summarizing my key points from ICCV’s Future of Real-Time SLAM Workshop, which was held on the last day of the conference (December 18th, 2015).

Today’s post contains a brief introduction to SLAM, a detailed description of what happened at the workshop (with summaries of all 7 talks), and some take-home messages from the Deep Learning-focused panel discussion at the end of the session.

Google has given some details on the near-miss experiences that its self-driving cars have had. Most, it turns out, were noted by the car’s own software, but a few required that the driver jump in without warning.

Chris Urmson, head of the driverless car program of Google’s parent company, Alphabet, said yesterday in a statement that there had been 341 incidents over 14 months. In 272 incidents, the self-driving system knew that it had failed and immediately alerted the driver; the humans took an average of less than a second to take the wheel.

As a biomedical informatics researcher, Nigam Shah spends his days using math to try to make sense of giant, unwieldy data sets. He’s used data mining to identify off-label drug use and to create widgets to add to electronic health records that may help doctors learn about best practices for the treatment of a disease in the absence of a randomized controlled trial. In one of his current projects, he’s using machine learning, a field of study that has helped develop self-driving cars and speech recognition software, to try to identify people with an under-diagnosed and potentially deadly genetic disorder.

The Data Science Development Program (DSDP) trains promising, recent graduates to become well-rounded data scientists for MassMutual. The DSDP is an intense, three-year program that combines academic coursework with the practical experience of working on data science projects for MassMutual. At the end of three years, this would be equivalent to earning a Master’s degree in data science with deep practical work experience. By the end of three years, members should be capable of leading their own data science projects for the company.

Deadline to Apply is Tuesday, March 1

Rice University is proud to host the 33rd annual meeting of the Society for Political Methodology. The conference will take place from July 21-23, 2016 at Rice University in the Bioscience Research Collaborative building.

The conference will include panel presentations of a single project and poster sessions for faculty members and graduate students that will provide ample opportunity for rich interaction and scholarly interchange.

Deadline to Apply is Monday, March 28

Following the code fraud committed by Volkswagen and in the midst of unregulated malware in advertising technology and ethically dubious use of algorithmic trading, the legal and ethical dimensions of data management and programming are fundamentally important components of comprehensive data scientific education.

This course begins by examining existing legal structures – copyright law, intellectual property, university technology office policies, patent law, patient protection (HIPAA) – in the US as they apply to data. International legal practices like France’s ‘right to be forgotten’ and internet censorship practices in countries including China and Turkey will also be investigated.

Earlier this month, IBM Press and Pearson have published my book titled: Getting Started with Data Science: Making Sense of Data with Analytics. You can download sample pages, including a complete chapter. There are 104 pages in the sample. You can also watch a brief interview about the book recorded earlier at the IBM Insight2015 Conference.

The very purpose of authoring this book was to rethink the way we have been teaching statistics and analytics to students and practitioners. It is no secret that most students required to take the mandatory stats course dislike it. I believe it has something to do with the way we have been teaching the subject than to do with the aptitude of our students.

One step in the development process for all software is the release. Some projects continually release, while others demarcate certain states of the codebase in a more formal, numbered way. Scientists leading the development of research software may be interested in how I manage my own software releases and encapsulate some reproducibility best practices along the way. A quick Google search turns up a plethora of “software release checklists.” This blog post summarizes mine.

This notebook explores a 3.9Gb CSV file containing NYC’s 311 complaints since 2003. It’s the most popular data set in NYC’s open data portal.

Today, we are proud to announce the public release of the largest-ever machine learning dataset to the research community. The dataset stands at a massive ~110B events (13.5TB uncompressed) of anonymized user-news item interaction data, collected by recording the user-news item interactions of about 20M users from February 2015 to May 2015.