NYU Data Science newsletter – February 17, 2016

GROUP CURATION: N/A

Discovering the visual features and representations used by the brain to recognize objects is a central problem in the study of vision. Recent successes in computational models of visual recognition naturally raise the question: Do computer systems and the human brain use similar or different computations? We show by combining a novel method (minimal images) and simulations that the human recognition system uses features and learning processes, which are critical for recognition, but are not used by current models. The study uses a “phase transition” phenomenon in minimal images, in which minor changes to the image have a drastic effect on its recognition. The results show fundamental limitations of current approaches and suggest directions to produce more realistic and better-performing models.

Along with the incredibly important task of writing research papers come the annoying conventions that the paper requires. Columns, charts, sourcing the charts and so much more, and these are not simple to navigate on a Word or Google Doc. This is precisely why Authorea was made. Authorea combines all the necessary features from your everyday word processors and adds all the tools you need for collaborative research articles.

Demis Hassabis has a modest demeanour and an unassuming countenance, but he is deadly serious when he tells me he is on a mission to “solve intelligence, and then use that to solve everything else”. Coming from almost anyone else, the statement would be laughable; from him, not so much. Hassabis is the 39-year-old former chess master and video-games designer whose artificial intelligence research start-up, DeepMind, was bought by Google in 2014 for a reported $625 million. He is the son of immigrants, attended a state comprehensive in Finchley and holds degrees from Cambridge and UCL in computer science and cognitive neuroscience. A “visionary” manager, according to those who work with him, Hassabis also reckons he has found a way to “make science research efficient” and says he is leading an “Apollo programme for the 21st century”. He’s the sort of normal-looking bloke you wouldn’t look twice at on the street, but Tim Berners-Lee once described him to me as one of the smartest human beings on the planet.

Artificial intelligence is already all around us, of course, every time we interrogate Siri or get a recommendation on Android. And in the short term, Google products will surely benefit from Hassabis’s research, even if improvements in personalisation, search, YouTube, and speech and facial recognition are not presented as “AI” as such.

In 2014, the former director of both the CIA and NSA proclaimed that “we kill people based on metadata.” Now, a new examination of previously published Snowden documents suggests that many of those people may have been innocent.

Last year, The Intercept published documents detailing the NSA’s SKYNET programme. According to the documents, SKYNET engages in mass surveillance of Pakistan’s mobile phone network, and then uses a machine learning algorithm on the cellular network metadata of 55 million people to try and rate each person’s likelihood of being a terrorist.

Patrick Ball—a data scientist and the director of research at the Human Rights Data Analysis Group—who has previously given expert testimony before war crimes tribunals, described the NSA’s methods as “ridiculously optimistic” and “completely bullshit.” A flaw in how the NSA trains SKYNET’s machine learning algorithm to analyse cellular metadata, Ball told Ars, makes the results scientifically unsound.

The other day, I was introduced to Jason at Clara Labs, a startup whose product is an AI personal assistant to schedule meetings. As it so happens, I use x.ai, a startup whose product is also an AI personal assistant to schedule meetings.

We wanted to meet for coffee, so I decided to let our AIs work it out.

What follows is the full correspondence between me, Jason, and our AIs Amy and Clara.

Recently we announced our efforts to make Yahoo a more focused company. This focus will let us accelerate the pace of innovation to make our products even better. We saw these changes as an opportunity to better align our research efforts, while preserving Yahoo’s culture of exploration and inquiry. As a result, we are reorganizing Yahoo Labs and moving forward with a new approach to research at Yahoo.

Today we’re announcing the expansion of Google Ideas, Google’s think tank, as a technology incubator called Jigsaw. The team’s mission is to use technology to tackle the toughest geopolitical challenges, from countering violent extremism to thwarting online censorship to mitigating the threats associated with digital attacks.

Also:

X, formerly Google[x], describes its process: The Secret to Moonshots? Killing Our Projects (Medium, Backchannel, Astro Teller; February 15)

Yahoo restructured its research organization: Yahoo’s New Research Model (Yahoo Research blog, Yoelle Maarek; February 16)

… At X, formerly called Google[x], you’ll find an aerospace engineer working alongside a fashion designer and former military ops commanders brainstorming with laser experts. These engineers, makers, and inventors dream up new technologies that we hope can lead to profound improvements in the way we live. We use the word moonshot to remind us to keep our visions big. To keep dreaming. And we use the word factory to remind us to have concrete plans for giving each vision its best chance to flourish.

But I have a secret for you. The Silicon Valley hype machine has created this myth of visionaries who effortlessly build the future. Don’t believe the hype. The moonshot factory is a messy place.

As humans, we have invested in the development of computers to automate the world around us in order to increase efficiency and the reach of human understanding. While computers can seamlessly make a large number of computations very quickly, the machines are little competition for the the biological machines we find in nature, like the brain. A developing research area in computer science is focused on better understanding biology and using the insights for algorithm development.

Following this path, Carnegie Mellon researcher Tai-Sing Lee, a professor in the Computer Science Department and the Center for the Neural Basis of Cognition (CNBC), will attempt to reverse-engineer the brain in order to try and reveal its learning methods and apply them to advancing machine learning algorithms. The project will span five years and will cost roughly $12 million.

In 2012, IBM Watson went to medical school. So said The New York Times, announcing that the tech giant’s artificially intelligent question-and-answer machine had begun a “stint as a medical student” at the Cleveland Clinic Lerner College of Medicine.

This was just a metaphor. Clinicians were helping IBM train Watson for use in medical research. But as metaphors go, it wasn’t a very good one. Three years later, our artificially intelligent machines can’t even pass an eighth-grade science test, much less go to medical school.

So says Oren Etzioni, a professor of computer science at the University of Washington and the executive director of the Allen Institute for Artificial Intelligence, the AI think-tank funded by Microsoft co-founder Paul Allen. Etzioni and the non-for-profit Allen Institute recently ran a contest, inviting nearly 800 teams of researchers to build AI systems that could take an eighth grade science test, and today, the Institute released the results: The top performers successfully answered about 60 percent of the questions. In other words, they flunked.

One of the sessions that I attended at the 2015 Moore-Sloan Data Science Environment Summit was titled “Isn’t Statistics Part of Data Science?” It is a niggling question I often consider, especially given how few statisticians there are at BIDS. A group of about forty people from statistics, computer science, and applied domains convened to discuss differences in practice and culture that divide statistics from data science. I am an applied statistician and a fellow at BIDS straddling these two worlds. I find it difficult to identify the line dividing these roles.

Wow us with your best data! And when we say “wow,” we mean just that: Make us laugh, make us cry, make us gasp with delight at the stunning discoveries and probing insights you can bring to life with data visualization. All entries should be in video format, and all videos should last no more than 90 seconds. Other than that, no holds barred. You can narrate, animate, or even act out your data points to tell us your bigger story.

Deadline for submissions is Friday, April 15.

Machine learning powers many Google product features, from speech recognition in the Google app to Smart Reply in Inbox to search in Google Photos. While decades of experience have enabled the software industry to establish best practices for building and supporting products, doing so for services based upon machine learning introduces new and interesting challenges.

Today, we announce the release of TensorFlow Serving, designed to address some of these challenges.