Data Science newsletter – April 29, 2019

Newsletter features journalism, research papers, events, tools/software, and jobs for April 29, 2019

GROUP CURATION: N/A

 
 
Data Science News



Northwestern Mutual invests in AI-based medical records startup

Milwaukee Business Journal, Nick Williams


from

Northwestern Mutual is investing up to $85,000 in a startup that’s developing an artificial intelligence-based platform for extracting data from electronic medical records.

Because some electrical medical documents are difficult to review, Pythonic AI, a new company co-founded by Baoqiang Cao and Matt Younkle, developed a machine learning platform to pull key words from electronic documents needed in underwriting insurance policies or evaluating claims.

Pythonic AI is receiving the investment as part of winning Northwestern Mutual’s Reverse Pitch MKE, a competition where the company invites entrepreneurs and startups to pitch tech solutions for problems presented by the Milwaukee-based life insurance and financial services giant. One of the challenges was finding a solution for streamlining the medical review process.


Your lunch is watching you

Axios, Kaveh Waddell


from

One unexpected byproduct of the robotization of food — an accelerating trend I reported on last week — is an explosion of data about eaters’ habits and preferences.

Why it matters: Companies often use this information to personalize food or ads to individual preferences. But seemingly trivial information about what and when you eat is also a gold mine that companies share with other interested parties — like your employer.

The same tradeoff at the center of the internet — personal information for convenience — is at play with trendy new food robots. When that data reveals something about employee work habits, their bosses get very interested.


How Recommendation Algorithms Run the World

WIRED, Business, Zeynep Tufeckci


from

What should you watch? What should you read? What’s news? What’s trending? Wherever you go online, companies have come up with very particular, imperfect ways of answering these questions. Everywhere you look, recommendation engines offer striking examples of how values and judgments become embedded in algorithms and how algorithms can be gamed by strategic actors.

Consider a common, seemingly straightforward method of making suggestions: a recommendation based on what people “like you” have read, watched, or shopped for. What exactly is a person like me? Which dimension of me? Is it someone of the same age, gender, race, or location? Do they share my interests? My eye color? My height? Or is their resemblance to me determined by a whole mess of “big data” (aka surveillance) crunched by a machine-learning algorithm?


Generation Z is stressed, depressed and exam-obsessed – For most youngsters getting good grades is a bigger worry than drinking or unplanned pregnancies

The Economist


from


How does it feel to be watched at work all the time?

BBC News, Padraig Belton


from

… Ben Waber, chief executive of Humanyze, a Boston workplace analytics company, says it gives firms the ability to assess how their staff are performing and interacting, which can be good for the firm but also good for employees themselves.

His company gathers “data exhaust” left by employees’ email and instant messaging apps, and uses name badges equipped with radio-frequency identification (RFID) devices and microphones.

These can check how much time you spend talking, your volume and tone of voice, even if you dominate conversations. While this may sound intrusive – not to say creepy – proponents argue that it can also protect employees against bullying and sexual harassment.

Humanyze calls these badges “Fitbit for your career”.


At Walmart, using AI to watch the store

Yahoo News, Associated Press, Anne D'Innocenzio


from

Inside one of Walmart’s busiest Neighborhood Market grocery stores, high resolution cameras suspended from the ceiling point to a table of bananas. They can tell how ripe the bananas are from their color.

When a banana starts to bruise, the cameras send an alert to a worker. Normally, that task would have relied on the subjective assessment of a human, who likely doesn’t have time to inspect every piece of fruit.

Welcome to Walmart’s Intelligent Retail Lab — the retail giant’s biggest attempt to digitize the physical store.


AKQA says it used AI to invent a new sport called Speedgate

TechCrunch, Anthony Ha


from

At TechCrunch, we write about AI all the time, whether the technology is being used to write books, make films or create a better McDonald’s drive-thru. But here’s one we haven’t heard before: using AI to invent a new sport.

The sport in question is Speedgate, and it was developed by AKQA. Creative Director Whitney Jenkins explained that the digital agency wanted to do something “really ambitious” for Design Week Portland, and given the team’s work with Nike (and its general “love of sports or athleticism”), it made sense to ask: “What if we invented the next basketball, the next football?”

To do that, AKQA says it used an existing recurrent neural network architecture, feeding it data about 400 sports, which were then used to generate sports concepts and rules.


Facebook’s Email-Harvesting Practice Is Under Investigation in N.Y.

Bloomberg Technology, Erik Larson


from

New York’s attorney general is opening an investigation into Facebook Inc.’s unauthorized collection of 1.5 million users’ email contacts without their permission.

The email harvest may have exposed hundreds of millions of people to targeted advertising by the embattled social-media company, New York Attorney General Letitia James said Thursday in a statement.


Canada accuses Facebook of breaking privacy laws

TheHill, Harper Neidig


from

Canada’s privacy watchdog is accusing Facebook of violating the country’s privacy laws in its handling of the Cambridge Analytica data scandal and vowing to take the social media giant to court.

Canada’s Office of the Privacy Commissioner (OPC) released its “troubling” findings on Thursday after a yearlong probe into Facebook’s privacy practices.


[1904.05234] Flash Boys 2.0: Frontrunning, Transaction Reordering, and Consensus Instability in Decentralized Exchanges

arXiv, Computer Science > Cryptography and Security; Philip Daian, Steven Goldfeder, Tyler Kell, Yunqi Li, Xueyuan Zhao, Iddo Bentov, Lorenz Breidenbach, Ari Juels


from

Blockchains, and specifically smart contracts, have promised to create fair and transparent trading ecosystems.
Unfortunately, we show that this promise has not been met. We document and quantify the widespread and rising deployment of arbitrage bots in blockchain systems, specifically in decentralized exchanges (or “DEXes”). Like high-frequency traders on Wall Street, these bots exploit inefficiencies in DEXes, paying high transaction fees and optimizing network latency to frontrun, i.e., anticipate and exploit, ordinary users’ DEX trades.
We study the breadth of DEX arbitrage bots in a subset of transactions that yield quantifiable revenue to these bots. We also study bots’ profit-making strategies, with a focus on blockchain-specific elements. We observe bots engage in what we call priority gas auctions (PGAs), competitively bidding up transaction fees in order to obtain priority ordering, i.e., early block position and execution, for their transactions.


Library launches initiative to boost data science expertise, services at UC Berkeley

UC Berkeley Library News, Virgie Hoban


from

In step with the explosion of data science across campus — including a new division, major, and undergraduate course that is the fastest-growing in university history — the Library has launched the UC Berkeley Library Data Initiatives Plan. Developed over years of conversations with librarians and campus partners, the plan is a multifaceted strategy for supporting Berkeley’s changing research landscape.

Data are, basically, blocks of information — numbers, survey results, images, map coordinates, and much more — that can be strung together and sorted at scale to answer a larger research question. With fine-tuned algorithms and the right computing power, data scientists can examine trends to explore issues ranging from humankind’s impact on climate change to the morphing of languages over time.

“With the right ideas and framework, you can do work that would have taken eons or billions of dollars within minutes or hours,” says Suen, who is helping the Library plan stations for data science consultation at the Center for Connected Learning, the new vision for Moffitt Library. “That’s the potential we have right now.”


As Georgia Looks to Expand Its Workforce, a New Kind of College Degree Aims to Establish Atlanta as the Financial Technology Capital of America

The 74, Aleksandra Appleton


from

Silicon Valley, Seattle and other major innovation hubs may be better known for their ability to attract talent in technology, but a state traditionally revered for its peaches and peanuts is introducing a big initiative to grow skilled tech workers right at home.

Georgia wants to position its college graduates at the forefront of the financial technology boom — by way of a new degree that offers highly specialized coursework and a foot in the door at major fintech companies.

The nexus degree in fintech from the University System of Georgia debuts this spring and entails 18 hours of coursework — approximately one year of study — in technical areas like payment transactions. At least six of those hours are spent in the field working for one of the financial technology companies that call Georgia home. The name of the degree — “nexus” — refers to this connection between learning in the classroom and learning on the job.


Pentagon’s Independent Research Group, the Jasons, Set to Disband

Gizmodo, Matt Novak


from

The Jason Group, an independent panel of academics who have advised the Pentagon for the past 59 years, will likely disband on April 30. The group had hoped to get a one-year extension to continue its work.

The Jasons was founded in 1960 as a scientific advisory panel that helped the Pentagon solve some of the most complex problems facing the U.S. military. The early days of the Jasons focused primarily on physics problems, but over the last six decades, the panel’s roughly 50 members have expanded their research to include studies on topics like artificial intelligence, health care, and climate change.

The Jason contract is managed by the MITRE Corporation, which allowed the group’s contract with the Department of Defense to expire on March 31, 2019. The Jasons advise other agencies like the Department of Energy, but without MITRE’s sponsorship, the group will have to dissolve completely and end all current studies for the DOE and other agencies by April 30.


New Turmoil Over Predicting the Effects of Genes

Quanta Magazine, Jordana Cepelewicz


from

over the past two decades experts have come up with robust statistical techniques to address the issue, using data collected from thousands of individuals. This approach has become particularly prevalent in human genetics, as researchers hope to predict, say, someone’s risk for a disease based on their genome. Some groups have even used these methods to probe how natural selection might have led to observed differences in height (and other traits) among populations. The findings generated further excitement about the potential applications in medicine and evolutionary biology for GWAS.

But now, two results published last month have cast doubt on those findings, and have illustrated that problems with interpretations of GWAS results are far more pervasive than anyone realized. The work has implications for how scientists think about the interactions between genetic and environmental effects. It also “raise[s] the ghosts of the possibility that we overestimate … how important genetics is in contributing to differences between people,” said Rasmus Nielsen, a biologist at the University of California, Berkeley.


The Terrifying Potential of the 5G Network

The New Yorker, Sue Halpern


from

A totally connected world will also be especially susceptible to cyberattacks. Even before the introduction of 5G networks, hackers have breached the control center of a municipal dam system, stopped an Internet-connected car as it travelled down an interstate, and sabotaged home appliances. Ransomware, malware, crypto-jacking, identity theft, and data breaches have become so common that more Americans are afraid of cybercrime than they are of becoming a victim of violent crime. Adding more devices to the online universe is destined to create more opportunities for disruption. “5G is not just for refrigerators,” Spalding said. “It’s farm implements, it’s airplanes, it’s all kinds of different things that can actually kill people or that allow someone to reach into the network and direct those things to do what they want them to do. It’s a completely different threat that we’ve never experienced before.”

Spalding’s solution, he told me, was to build the 5G network from scratch, incorporating cyber defenses into its design. Because this would be a massive undertaking, he initially suggested that one option would be for the federal government to pay for it and, essentially, rent it out to the telecom companies. But he had scrapped that idea.

 
Events



The AI Summit, London 2019

AI Business


from

London, England June 12-13. “New for the 4th Annual, edition, AI will take centre stage to TechXLR8: Europe’s largest tech expo focused on Accelerating Business Transformation with Technology. Co-located with 5 world leading enterprise technology events on IoT, Cloud, Quantum Computing, as well as being supported by the world’s great change-makers, The AI Summit gives you the depth and breadth to light up your technology roadmap and bring your business into the 4th Industrial Revolution powered by AI.” [$$$$]


Xconomy: Announcing Net@50: The Roots and Future of the Internet

Xconomy, World Frontiers Forum


from

Cambridge, MA July 16. “Net@50: The Roots and Future of the Internet, a unique two-part event to be held at the MIT Media Lab and Café ArtScience … we will pay tribute to the internet pioneers and also look ahead to what the next 50 years might bring, for both better and worse.” [$$$]

 
Deadlines



Startup Showcase: Data Tech 2019

Bloomington, MN May 30, Normandale College. “This event is our most technical. It will include a Startup Showcase session with pitches from analytics, AI, machine learning, and other emerging data technology startups.”
 
Tools & Resources



Why You Need a Data Taxonomy

Neilsen, Insights


from

One of the tools for digital marketers is multi-touch attribution. Multi-touch attribution takes all of the touchpoints on the consumer journey into consideration and assigns fractional credit to each so that a marketer can see how much influence each channel has on a sale. This gives greater insight than last-touch attribution, which assigns 100% of the credit to the last touchpoint, leaving the marketer without a view of the influence of other channels.

What do we mean by that? We mean that multi-touch attribution allows marketers to know which points on a consumer’s path to purchase contributed to a sale, rather than assuming the last touchpoint triggered it.


A Recipe for Training Neural Networks

Andrej Karparthy


from

Some few weeks ago I posted a tweet on “the most common neural net mistakes”, listing a few common gotchas related to training neural nets. The tweet got quite a bit more engagement than I anticipated (including a webinar :)). Clearly, a lot of people have personally encountered the large gap between “here is how a convolutional layer works” and “our convnet achieves state of the art results”.

So I thought it could be fun to brush off my dusty blog to expand my tweet to the long form that this topic deserves. However, instead of going into an enumeration of more common errors or fleshing them out, I wanted to dig a bit deeper and talk about how one can avoid making these errors altogether (or fix them very fast). The trick to doing so is to follow a certain process, which as far as I can tell is not very often documented. Let’s start with two important observations that motivate it.


What’s Next for Graphs: Neo4j for Google Cloud

Neo4j, Emil Eifrem


from

“We’ve formally announced a new strategic partnership with Google Cloud that delivers Neo4j as a fully managed service deeply integrated with the Google Cloud Platform.


This proposal introduces callables to Swift.

GitHub – apple


from

Callables are values that define function-like behavior and can be applied using function application syntax.

 
Careers


Postdocs

Postdoctoral Positions



Johns Hopkins University, Center for Language and Speech Processing; Baltimore, MD
Full-time, non-tenured academic positions

Research Data Specialist



Fred Hutchinson Cancer Research Center; Seattle, WA

Leave a Comment

Your email address will not be published.