NYU Data Science newsletter – September 29, 2015

GROUP CURATION: N/A

Human Connectome Project finds surprising correlations between brain architecture and behavioural or demographic influences.

Christine Borgman, who studies how research information is retrieved, processed, curated and conveyed, is in the right place at the right time — when the demand for data by researchers and scholars in many different disciplines is greater than ever before.

But sharing data isn’t as simple as it sounds.

Neural networks are generating a lot of excitement, as they are quickly proving to be a promising and practical form of machine intelligence. At Fast Forward Labs, we just finished a project researching and building systems that use neural networks for image analysis, as shown in our toy application Pictograph. Our companion deep learning report explains this technology in depth and explores applications and opportunities across industries.

When Jacob Aron helped judge an artificial intelligence contest, the entrants did not interview well. Better to judge face recognition or even poker skills.

… In recent years, however, the acceleration predicted by Moore’s Law has slipped. Chip speeds stopped increasing almost a decade ago, the time between new generations is stretching out, and the cost of individual transistors has plateaued.

Technologists now believe that new generations of chips will come more slowly, perhaps every two and a half to three years. And by the middle of the next decade, they fear, there could be a reckoning, when the laws of physics dictate that transistors, by then composed of just a handful of molecules, will not function reliably. Then Moore’s Law will come to an end, unless a new technological breakthrough occurs.

… Stylit is an Israeli startup that brings together human stylists and a machine learning algorithm to help users pick out clothes. Co-CEO Assif Versano says Stylit is “more data company than personal styling service.” As with most online styling services, the process begins with an online profile. Then, users receive weekly curated “looks” on the Stylit website and can order any items that they like.

As a teenager in the early 2000s, London artist Alexander Taylor and his friends shot videos on their Motorola Razr-era cellphones. They swapped the movies they made with one another in the already largely forgotten .3gp file format.

“It used to be quite a thing for all my friends to create these 3G videos,” says Taylor, now 24.

Years later, Taylor discovered a treasure trove of amateur feature-phone videos from around the world can be found on YouTube by searching for that .3gp extension. Many of them, he says, appear to have been bulk-uploaded from their creators’ phones or hard drives, with the filenames, complete with extension, becoming the YouTube titles.

Most of these videos have been watched only rarely—but that may soon change.

As with many aspects of society, human nature shapes interactions in science research. When we consider “data sharing,” the likely response is probably a shrug. We’ve all been there. Group work and competition at its finest. The increasingly competitive environment for grant funding, and the ‘publish or perish’ attitude promotes the “mine, mine, mine” attitude among scientists. To focus on the issue of overcoming career-protecting objections to data sharing however, we can focus on several trends.

Continuum Analytics has a mission to help people around the world discover, produce, share, and collaborate better—facilitating the connection of expertise and data. “People are amassing larger and larger amounts of data,” explains Director of Sales Engineering and Implementation Duane Lawrence. “We’re helping them to work with that data.” How, you might be wondering? By building technologies that engage communities and collect information, Continuum provides analysts with a better medium to communicate their decisions—and focus their stakeholders on the details that matter most.

If there’s one key value that defines Continuum’s culture, it’s collaboration.

A little over a year ago, we blogged about jugyokenkyu, or “lesson study”, a bucket of practices that Japanese teachers use to hone their craft, from observing each other at work to discussing the lesson afterward to studying curriculum materials with colleagues. Getting the Software Carpentry Foundation up and running almost immediately pushed that aside, but now that the SCF is up and running, it’s time to return to the subject. Discussion of how teaching practices are transferred is part of that; so are two other developments this week.

Train…Test…Deploy…

Building cognitive Question and Answer applications is now as easy as 1 – 2 – 3 with the IBM Watson Natural Language Classifier and Watson Dialog services. These two APIs have commonly been combined in enterprise applications but we are now making them available to developers as part of the Watson Developer Cloud on the IBM Bluemix platform. With these services you can embed natural language processing, conversation and deep learning to build engaging cognitive applications quickly and easily.

A Kudu cluster stores tables that look just like tables you’re used to from relational (SQL) databases. A table can be as simple as an binary key and value, or as complex as a few hundred different strongly-typed attributes.

Just like SQL, every table has a PRIMARY KEY made up of one or more columns. This might be a single column like a unique user identifier, or a compound key such as a (host, metric, timestamp) tuple for a machine time series database. Rows can be efficiently read, updated, or deleted by their primary key.

Kudu’s simple data model makes it breeze to port legacy applications or build new ones.