Data Science newsletter – November 3, 2016

GROUP CURATION: N/A

NV) It has been 12 years since the publication of ON Intelligence from Jeff Hawkins. In this book the founder of Numenta was setting the vision for AI. Now everybody talks about AI, as it transforms our life. Did this book come too early? What was the missing chain? Why did it take twelve years for AI to mature?

AM) Wow! Has it really been that long?! I first read On Intelligence as a neuroscience grad student and found it to offer a refreshing perspective that certainly helped steer my career. In On Intelligence, Jeff proposes a simple and straightforward theory of intelligence in terms of the structure and circuitry of the neocortex. Jeff also makes the claim that the approaches to AI at the time were not on track to making computers intelligent. Specifically, Jeff challenges the commonly held belief that computers will be intelligent when they are powerful enough. The timing of his claim is notable in that in the time since publishing On Intelligence, there has been a renewed interest in applying artificial neural networks catalyzed, in part, by vast improvements in computing performance yielding some impressive results and useful models. It has been interesting watching the fields of Artificial Intelligence and Neuroscience converge. There’s still so much to learn from the brain and we’ve only barely scratched the surface in applying what we’ve learned. Despite its progress, 12 years later, the field of AI still feels very much in its infancy with lots of room to mature.

‘Things are not what they used to be.’ How often those in the older generation use this phrase to scold the morals, attitudes and behaviour of younger rivals. And yet, how often do the same people, often in positions of power and responsibility, deny the changes in circumstance that newer generations complain about with justification. So, let’s be clear: young scientists today face a harsher, more competitive, stricter, more dispiriting workplace than their bosses and senior colleagues did at the same stages of their own careers. Things are simply not the same as they were back in the day. They are more difficult. In a special issue, Nature examines the problems and the possible fixes.

The research community — from individual scientists to institutions and funders — must respond. Much has been written, in these pages and elsewhere, about the glut of PhD students and the insecurity of the postdoc years. It is hard, and getting harder, to get a foot in the research door. Which makes it all the more galling that those who rise to the level of principal investigator, perhaps with an opportunity to build their own lab or group, do not receive the focused support they need to flourish, to sustain their hard-won position and convert it to career success. Universities, funders, senior figures: your principal investigators need you to recognize their struggle and introduce concrete changes to help them.

The US Securities and Exchange Commission’s (SEC) Kara Stein has renewed calls for the creation of an Office of Data Strategy to overhaul the SEC’s “ad hoc” data collection and analysis strategies going forward.

Speaking at the the Big Data in Finance Conference in Michigan, commissioner Stein explained: “For some time, I have asked that the SEC develop an executive team responsible for creating and overseeing such an office.”

Why is it so rare for customer insight (CI) functions to have a seat at the table when key decisions are being made? Executives in consumer-facing companies know that understanding customers’ motivations and anticipating their behavior can accelerate and amplify growth. Yet most organizations struggle to integrate CI into their strategic decision making and core processes.

At Stanford, scholars and students are looking for creative ways to make cities better places for people to live and thrive – places that offer quality and affordable housing, desirable public spaces, robust transportation systems, healthy air and water, and economic promise for all.

Great cities foster human relationships and social inclusion, said Deland Chan, a lecturer in urban studies and co-founder of Stanford’s Human Cities initiative: “This is as much about the process of decision-making in designing our cities as it is about technological solutions.” [video, 2:28]

Today, a new student arriving in my group finds all of our research code in tidy repositories, where every change is recorded automatically. Version control is our essential technology for record keeping and collaboration. Whenever we publish a paper, we create a “reproducibility package,” deposited online, which includes the data sets and all the code that is needed to recreate the analyses and figures. These are the practices that work for us as computational scientists, but the principles behind them apply regardless of discipline.

It takes new students some time to learn how to work to these standards, but we have documentation and training materials to make it as painless as possible. My students don’t resent investing their time in this. They know that practices like ours are crucial for the integrity of the scientific endeavor.

In the arena of cybersecurity, an increasing number of companies are incorporating artificial intelligence in their new products. A growing dependence on computers demands the creation of a smart and autonomous security system. Recently, SparkCognition revealed one of the first “cognitive” antivirus systems, DeepArmor. Powered by artificial intelligence, the system protects networks from new threats using heuristics, neural networking, natural language processing, and data science. These techniques fuel the identification and removal of never-before-seen malicious software.

Partnering with NASA and the University of Wisconsin-Madison, the Wisconsin Department of Natural Resources (DNR) has launched a wildlife observation program called Snapshot Wisconsin that will be one of the largest trail camera projects ever deployed.

The DNR will set up 4,000 to 5,000 motion-sensor cameras throughout the state to capture photos of the state’s wildlife, including deer, bears, elk, coyotes, bobcats, badgers and whatever else triggers the camera shutter.

A revolutionary technology is just that: revolutionary. To appreciate its potential, we need to break out of thought patterns that were formed when the technology did not exist. We need to rethink designs that are built on its absence.

To illustrate this, let me give you an example from a domain I’m intimately familiar with: video game development. The video game industry is arguably the largest of the entertainment industries, and definitely the fastest growing. It is also inherently high-tech; games rely on advances in processing, graphics, and interfaces, among other technologies. Many of its employees are young and infatuated with technology. The game industry should be receptive and eager for advances in AI.

Many game designers and developers, however, think some new AI methods are not needed in their games, for a whole host of reasons.

Two Sigma, in partnership with Cornell Tech, today announced the public launch of Halite, a programming game in which players code bots that compete head-to-head to overtake a virtual grid. Halite provides a fun way to learn and apply AI, machine learning, and other advanced algorithms in a collaborative, competitive game setting by writing smart bots. Designed for coding enthusiasts of all levels of experience, Halite creates an engaging game environment to learn, write, and visualize your code in action. Users can track their own bots as well as their competitors’ progress—either globally or within private groups at a company, school, or club—by viewing a real-time leaderboard. The game will be released for a three-month competition, and the success of each bot will be correlated with the creativity and sophistication of its code.

The Trending product has repeatedly promoted false news, and experts say it may get worse as the company scales Trending internationally.

Netatmo is launching Presence — a smart outdoor security camera that will send you an alert if someone is loitering around your house. Presence is available now in the U.S. market for $300.

The product is one more piece of the smart home, which is expected to become a $71 billion market by 2018, according to Jupiter Research. Paris-based Netatmo unveiled Presence earlier this year, saying that the camera uses a deep-learning algorithm to detect people, cars, and animals that are within view.

Pittsburgh residents will now have greater insight into public records around crime, code violations, noise complaints and more, thanks to new open data portal Burgh’s Eye View, Trib Live reports. The app allows viewers to see arrests, police blotter entries, city facilities and other factors, including the gamut of 311 requests, mapped onto where they occurred in the city.

Ninety percent of open data experts interviewed in a new report believe the standardization and publication of government data have improved over the last few years of the Obama administration.

The report, released jointly today by the Data Foundation, an open data research organization based in Washington, D.C., and consulting giant Grant Thornton, includes a history of U.S. open data efforts and detailed feedback from more than 40 data transparency experts within and outside government.

The Gordon and Betty Moore Foundation announced the first five of what will eventually be 50 Moore Inventor Fellows. Each fellow will receive a total of US $825,000 over three years to drive their invention forward, including $50,000 per year from their institution. All told, the Moore Foundation plans to invest $34 million.

“We are investing in promising scientist-problem solvers with a passion for inventing—like Gordon Moore himself,” said Harvey V. Fineberg, president of the Gordon and Betty Moore Foundation, in a press release.

… 3. What do you wish you knew earlier about being a data scientist?

I was doing what is now called data science at Amazon.com in 1997.The term wasn’t even coined until 2008 (by Jeff Hammerbacher and DJ Patil). It’s hard to be much earlier. As for what I wish, I mostly wish I had the powerful tools we have now back then; today is a wonderland of data, tools, and computation. It’s a great time to be a data scientist.

… I got to UCLA in 1969. I immediately got interested in statistical decision theory and decision analysis. It took me ten years to get into what I’ve been doing since, namely, automated decision making. The only group that was into this challenge was Ron Howard’s, in management, not in computer science.

In the late ’70s and early ’80s, everybody in AI was working on expert systems for all kinds of applications, from medical diagnosis to mineral explorations. The idea was that, wherever you pay a professional, often called “expert,” you can emulate that professional on a computer. By interviewing the professional, you can extract the basic rules by which he or she operates and, once you have a computer full of rules, you have an engine that can activate the rules in response to the evidence observed, and this will tell you, for example, where to dig for oil or what medical test to conduct next. [video, 32:02]

“Exhibitions from our founding in 1929 to the present are available online (e.g. GitHub). These pages are updated continually.”

“A curated list of awesome TensorFlow experiments, libraries, and projects. Inspired by awesome-machine-learning.”

Data science continues to generate excitement and yet real-world results can often disappoint business stakeholders. How can we mitigate risk and ensure results match expectations? Working as a technical data scientist at the interface between R&D and commercial operations has given me an insight into the traps that lie in our path. I present a personal view on the most common failure modes of data science projects.

Large computer-understandable proofs consist of millions of intermediate
logical steps. The vast majority of such steps originate from manually
selected and manually guided heuristics applied to intermediate goals.
So far, machine learning has generally not been used to filter or
generate these steps. In this paper, we introduce a new dataset based on
Higher-Order Logic (HOL) proofs, for the purpose of developing new
machine learning-based theorem-proving strategies. We make this dataset
publicly available under the BSD license.