Data Science newsletter – October 8, 2019

GROUP CURATION: N/A

Companies are selling reports about a wide range of physical, cognitive, and behavioral traits to consumers based on their genomic data, but such tests have a number of limitations.

It can pick out shoplifters, international criminals and lost children in seconds. But as the cameras proliferate, who’s watching the watchers?

Gordon’s wine bar is reached through a discreet side-door, a few paces from the slipstream of London theatregoers and suited professionals powering towards their evening train. A steep staircase plunges visitors into a dimly lit cavern, lined with dusty champagne bottles and faded newspaper clippings, which appears to have had only minor refurbishment since it opened in 1890. “If Miss Havisham was in the licensing trade,” an Evening Standard review once suggested, “this could have been the result.”

The bar’s Dickensian gloom is a selling point for people embarking on affairs, and actors or politicians wanting a quiet drink – but also for pickpockets.

As we teeter on the brink of another technological revolution—the artificial intelligence revolution—worry is growing that it might be our last. The fear is that the intelligence of machines will soon match or even exceed that of humans. They could turn against us and replace us as the dominant “life” form on earth. Our creations would become our overlords—or perhaps wipe us out altogether. Such dramatic scenarios, exciting though they might be to imagine, reflect a misunderstanding of AI. And they distract from the more mundane but far more likely risks posed by the technology in the near future, as well as from its most exciting benefits.

The fifth International Conference on Computational Social Science (IC2S2) took place this past July in Amsterdam. While computational social science has existed for longer than IC2S2, it has gone through a revolution in the last ten years, due, in part, to the explosion of big data and scientists’ ability to analyze these large datasets. This version of computational social science, described ten years ago in a perspective by David Lazer and colleagues as “a computational social science… that leverages the capacity to collect and analyze data with an unprecedented breadth and depth and scale (1),” included research on almost every topic related to human behavior. This, to me, is what makes computational social science unique. Instead of being a discipline centered around a single branch of knowledge (such as neuroscience is centered around the brain), it is a discipline characterized by the methods and data used to answer questions that previously would have fallen into disparate disciplines.

This is readily apparent from the range of topics covered by both the keynote and parallel oral presentations. It would be easiest to describe the range by providing a list, but since nobody likes to read lists, I will instead discuss some of the questions that computational scientists are answering.

Facebook has been able to share some information with researchers, but providing them with more sensitive and detailed data without compromising user privacy proved technically more difficult than the project’s organizers expected.

Last month, the 8 charitable funders — which so far have provided a total of up to US$600,000 for the scheme, called the Social Media and Democracy Research Grants programme — gave Facebook until 30 September to provide the full data set or said they would begin winding up the programme. They say that it is impractical to allow researchers to keep bidding for cash while no one knows when the necessary data will become available. The programme’s structure — which included separate bodies to oversee grants and to provide access to the data — had also proven too complex, says Larry Kramer, president of one of the charities, the Hewlett Foundation in Menlo Park, California.

Google plans to move its downtown Portland office into the historic Meier & Frank Building next year, occupying two floors and making room for further growth.

Google has had a Portland office since 2010 when it bought an Oregon software company called Instantiations.

Plate Tectonics emerged in the late1960s when geologists realized that plates moving on Earth’s surface at fingernail-growth speeds side-swipe each other at some places (like California) and converge at others (like Japan). When they converge, one plate plunges down into Earth’s mantle under the other plate, but what happened to it deeper in the mantle remained a mystery for most of the 20th century. Like an ancient map labeled “here be dragons,” knowledge of the mantle remained skin-deep except for its major boundaries.

Now a marriage of improved computing power and new techniques to investigate Earth’s interior has enabled scientists to address some startling gaps in the original theory, like why there are earthquakes and other tectonic phenomena on continents thousands of miles from plate boundaries.

Digital mental health company SilverCloud Health has partnered with Microsoft Labs in Cambridge to explore how machine learning and artificial intelligence (AI) could be used to personalise mental healthcare.

UC Santa Barbara alumnus Ross Dowd ‘94 has spearheaded an upcoming humanities-based technological program — the Creative Computing Initiative — meant to prepare humanities majors for careers in an increasingly tech-driven world. … The initiative will consist of various programming, data-driven and technical courses designed specifically for humanities students to create an interdisciplinary curriculum for those in pursuit of a liberal arts degree.

With cybercriminals taking less and less time to break into corporate systems, enterprises will have to tap artificial intelligence and machine learning tools to bolster their ability to defend against attacks and beef up their network resilience.

France is poised to become the first European country to use facial recognition technology to give citizens a secure digital identity — whether they want it or not.

Saying it wants to make the state more efficient, President Emmanuel Macron’s government is pushing through plans to roll out an ID program, dubbed Alicem, in November, earlier than an initial Christmas target. The country’s data regulator says the program breaches the European rule of consent and a privacy group is challenging it in France’s highest administrative court. It took a hacker just over an hour to break into a “secure” government messaging app this year, raising concerns about the state’s security standards.

If Energy Secretary Rick Perry leaves the Trump administration by year’s end, as media reports say, he’ll leave a Department of Energy that’s far larger than when he arrived.

DOE’s annual budget for fiscal 2020 could approach $40 billion—a nearly 25% increase from when Perry took the helm in fiscal 2017. That’s an astonishing expansion of spending for a department that Republican presidential hopeful Perry had targeted for elimination before he comically forgot its name in a 2012 candidates’ debate.

Smart cities will provide more and more services, an increasing number of which they’ll be apt charge for. As urban IoT infrastructure expands, local businesses will naturally also take advantage of it. With all of that on the way, there’s an expectation that it will be useful for people to have access to a suitable electronic currency for small fees and minor purchases — for making micropayments. IOTA, a ledger-based technology designed specifically for the IoT, is being proposed for that purpose. … IOTA is based on a new distributed ledger, called Tangle, which overcomes the inefficiencies of the current blockchain design, introducing a new method of consensus in a decentralized peer-to-peer solution. This approach allows using the technology, for example, to transfer money without any commission to pay for anything from parking fees to a car wash to a sandwich.

Animal survival requires a functioning nervous system to develop during embryogenesis. Newborn neurons must assemble into circuits producing activity patterns capable of instructing behaviors. Elucidating how this process is coordinated requires new methods that follow maturation and activity of all cells across a developing circuit. We present an imaging method for comprehensively tracking neuron lineages, movements, molecular identities, and activity in the entire developing zebrafish spinal cord, from neurogenesis until the emergence of patterned activity instructing the earliest spontaneous motor behavior. We found that motoneurons are active first and form local patterned ensembles with neighboring neurons. These ensembles merge, synchronize globally after reaching a threshold size, and finally recruit commissural interneurons to orchestrate the left-right alternating patterns important for locomotion in vertebrates. Individual neurons undergo functional maturation stereotypically based on their birth time and anatomical origin. Our study provides a general strategy for reconstructing how functioning circuits emerge during embryogenesis. [full text]

New York, NY November 14, starting at 3 p.m. Panelists: Margo Anderson, danah boyd, Mark Hansen, Melva M. Miller. [free, registration required]

Ann Arbor, MI October 19, starting at 8:30 a.m. Research talks will be given by University of Michigan AI faculty: Ella Atkins, Satinder Singh Baveja, Joyce Chai, David Fouhey, Justin Johnson, and Edwin Olson. [sold out, waiting list available]

… Executive control must be paid its due. Many of life’s challenges require nothing more than this. A decision to ask for a raise at work starts with setting an appointment with your boss. You carefully phrase your request and outline your reasons. Or, you decide to add some romance to your life by asking that attractive person at the gym to meet for coffee. After some deliberation, you find an appropriately casual way to do so. Decisiveness works in these one-off events. We make our decision, steel our resolve, and muster our strength to follow through.

Other parts of our lives, however, are stubbornly resistant to executive control. And thinking every time we act would, in any case, be a highly inefficient way of conducting our lives. I’ll return to this later, but can you imagine trying to “make the decision” to go to the gym every single time you went? You’d be condemning yourself to rekindle the ardor of Day One every single day. You’d be forcing your mind to go through that exhausting process of engaging with all the reasons that you felt you should be going to the gym in the first place— and, because our minds are wonderfully, irrationally adversarial, you’d have to run through the reasons not to go, too. Each time. Every day. That’s how our minds work if we let them. You would constantly be in the throes of heavy mental lifting, with little time to think about anything else.

These are our habits—better suited to working automatically than to engage in the noisy, combative work of the debating chamber that usually accompanies our decision-making. What we’ll also see is that a whole lot of life is already contained in those automatic parts—the simple, assiduous parts of yourself that you can set to a task. What could be better than that for accomplishing important and long-term goals? Skip the debating chamber and get to work. That’s exactly what habits are for.

This blog post is about a model that has attracted a lot of attention in the past year: deep learning in the regime where the width – namely, the number of channels in convolutional filters, or the number of neurons in fully-connected internal layers – goes to infinity. At first glance this approach may seem hopeless for both practitioners and theorists: all the computing power in the world is insufficient to train an infinite network, and theorists already have their hands full trying to figure out finite ones. But in math/physics there is a tradition of deriving insights into questions by studying them in the infinite limit, and indeed here too the infinite limit becomes easier for theory.

“emcee is an MIT licensed pure-Python implementation of Goodman & Weare’s Affine Invariant Markov chain Monte Carlo (MCMC) Ensemble sampler and these pages will show you how to use it.” … “emcee has been used in quite a few projects in the astrophysical literature and it is being actively developed on GitHub.”