Data Science newsletter – October 3, 2018

GROUP CURATION: N/A

The company’s platform uses single B cell screening and advanced sequencing to unearth huge amounts of data on antibodies. It then uses machine learning and custom data visualization tools to narrow down the data into antibodies that could be used for a variety of therapeutic options, like antibody immunotherapy treatments for cancer.

As Hurricane Florence ground its way through North Carolina, it released what might politely be called an excrement storm. Massive hog farm manure pools washed a stew of dangerous bacteria and heavy metals into nearby waterways.
Satellite images of river outflows to the Atlantic Ocean in the wake of Hurricane Florence show water discolored by debris including pollutants spilled by hog farms.

More efficient oversight might have prevented some of the worst effects, but even in the best of times, state and federal environmental regulators are overextended and underfunded. Help is at hand, however, in the form of machine learning – training computers to automatically detect patterns in data – according to Stanford researchers.

Their study, published in Nature Sustainability, finds that machine learning techniques could catch two to seven times as many infractions as current approaches, and suggests far-reaching applications for public investments.

Those of us who worked to create the internet during the Cold War were driven by the deep belief that open communication and collaboration could change our world for the better. But that belief was not a guarantee. Today, the same technological advancements that have dramatically increased the ease of communicating and sharing information are being used for harm and hatred.

Online groups claiming freedom of speech proudly disseminate instructions for creating untraceable firearms with 3D printers. The Director of National Intelligence, Dan Coats, warns that the digital infrastructure serving this country is under attack by foreign agents. Facebook’s Mark Zuckerberg admits to feeling “a deep sense of responsibility” to fix disinformation and hate speech on the platform, which has been linked to deadly violence in several countries.

It would be a serious mistake to underestimate the scale of these threats, or the seriousness with which the tech community must work to create effective solutions. However, it would also be a mistake to think that technology is inherently to blame for the proliferation of hatred.

Last year, theoretical chemist and Princeton professor Salvatore Torquato had a hunch—what if prime numbers were modelled as atom-like particles? Would they create a pattern too?

According to recently released US federal contracting data, the Drug Enforcement Administration will be expanding the footprint of its nationwide surveillance network with the purchase of “multiple” trailer-mounted speed displays “to be retrofitted as mobile LPR [License Plate Reader] platforms.” The DEA is buying them from RU2 Systems Inc., a private Mesa, Arizona company. How much it’s spending on the signs has been redacted.

Two other, apparently related contracts, show that the DEA has hired a small machine shop in California, and another in Virginia, to conceal the readers within the signs. An RU2 representative said the company providing the LPR devices themselves is a Canadian firm called Genetec.

The DEA launched its National License Plate Reader Program in 2008; it was publicly revealed for the first time during a congressional hearing four years after that. The DEA’s most recent budget describes the program as “a federation of independent federal, state, local, and tribal law enforcement license plate readers linked into a cooperative system, designed to enhance the ability of law enforcement agencies to interdict drug traffickers, money launderers or other criminal activities on high drug and money trafficking corridors and other public roadways throughout the U.S.,” primarily along the southwest border region, and the country’s northeast and southeast corridors.

China is building a digital dictatorship to exert control over its 1.4 billion citizens. For some, “social credit” will bring privileges — for others, punishment.

With every President comes a new NASA administrator, and the current admin, Jim Bridenstine, has raised a number of eyebrows. The strongest reaction to Bridenstine’s appointment comes from his lack of a science background, though more recent reports say he has changed his mind on climate change and does believe humans are responsible and can curb the effects we’re having on the planet. Nevertheless, the immediate knee-jerk reaction I saw from the space community raised the question is of whether a scientist is really the right person to run NASA.

To answer this question, we have to start by taking a quick look at past administrators and their backgrounds. Head’s up — this is going to be a very cursory look but each Administrator’s name is linked to their bio if you want to read a little more.

Aviv Regev speaks with the urgent velocity of someone who has seen the world with an extraordinary new acuity, and can’t wait for you to hurry up and see it too. At a meeting of 460 international scientists gathered last week in San Francisco, the computational biologist bombarded her audience with a torrent of results from her lab at the Broad Institute of MIT and Harvard, where she is pioneering powerful new tools for understanding what we humans are really made of—and what makes us fall apart.

“Where do disease risk genes act?” she fired into the crowd. “Which molecular communications are being disrupted? Which cell programs are being changed? These are the next generation of questions we can now ask.”

For centuries, scientists like Regev have known that clues to our elemental humanity were hiding in the basic unit of life: the cell.

I’ve always believed the web is for everyone. That’s why I and others fight fiercely to protect it. The changes we’ve managed to bring have created a better and more connected world. But for all the good we’ve achieved, the web has evolved into an engine of inequity and division; swayed by powerful forces who use it for their own agendas.

Today, I believe we’ve reached a critical tipping point, and that powerful change for the better is possible — and necessary.

This is why I have, over recent years, been working with a few people at MIT and elsewhere to develop Solid, an open-source project to restore the power and agency of individuals on the web.

Why doesn’t the app provide safer walking directions to women, or indeed anybody else, at night? It knows where we are, it knows what time the sun sets. It knows the population density of the area, and crime statistics are geotagged and publicly available in most major cities. It can even see which areas people walk through with impunity during the day but avoid in the dark.

Other Google Maps blind spots, such as providing information on accessibility for those with limited mobility, are starting to see change. The presence of a single stair or roadside curb without a ramp could be prohibitive for users in a wheelchair. After years of campaigning by disability activist groups, in March this year Google launched an accessibility function that navigates people along routes with step-free access. But this required an intensive manual effort by humans, called “guides”, to answer questions within the app about accessibility of buildings, roads and Tube stations. This manual data layering is incredibly “information labour”-intensive, says Gina Neff, associate professor at the Oxford University Internet Institute.

Part of the reason Google doesn’t do this automatically is down to the evolution of the Maps’ function. Google Maps was originally designed for – and by – driving.

As competition for analytical minds in finance increases, one firm has taken a novel approach to unearth fresh talent: a global tournament for quants.

WorldQuant, a hedge fund in Connecticut, started the International Quant Championship this year. The competition attracted 11,000 applicants, from over 1,000 universities and more than 80 countries, to develop new types of trading algorithms. Up for grabs? Potential full-time positions or internships at the hedge fund.

This month, 47 finalists from 15 countries met in Singapore for the finals. An all-women team from the Moscow Institute of Physics and Technology was declared the overall winner, but all of the finalists were given the chance to interview for full-time and internship roles at the fund.

Ann Arbor, MI October 8-9, University of Michigan Rackham Building (915 E. Washington St.) [registration required]

Brooklyn, NY Saturday, November 3, starting at 9:30 a.m., Pratt Institute. “A one-day conference that brings together experts from the worlds of science, design, art, theater, and activism to examine new ways of communicating, working and collaborating.” [$$$]

SAGE Advice connects you, the researcher, with a data science expert with the right technical expertise. Via a 30- or 60-minute online video one-to-one consultation, the Advice expert will answer your questions and suggest solutions to help you move your research project forward. Cost is currently $50 for the 30-minute session and $100 for the hour-long version, but during this trial period the Advice team is offering 30-minute sessions at no cost.

Earlier this year, a new package called tabulizer was released in R, which allows you to automatically pull out tables and text from PDFs. Note, this package only works if the PDF’s text is highlightable (if it’s typed) — i.e. it won’t work for scanned-in PDFs, or image files converted to PDFs.