ML/DL

Machine learning (ML) and deep learning (DL) content from the past 24 hours.

Note to readers: Big Analytics will rebrand as ML/DL soon.

Good Reads

— Adrian Colyer summarizes a paper on deep symbolic reinforcement learning.

— Another excellent piece from Nicole Hemsoth, on Bitfusion, a startup that delivers software-defined supercomputing through virtualization and efficient use of accelerators.

— Davey Winder summarizes how advances in machine learning combined with new data from wearables help insurers predict risk and detect fraud.

— Jason Lim summarizes what leading ML and AI scientists say about the future.

White House Report on AI

— The White House publishes a report: Preparing for the Future of Artificial Intelligence. In Fortune, Barb Darrow summarizes. The report, which is 50 pages long, includes a paragraph on discrimination. Jordan Pearson thinks it’s all about racism.

Bias in Machine Learning

— Yvonne Baur asks: Is machine learning sexist? She recaps the Word2vec problem, reported in MIT Technology Review and an academic paper; a neural network trained on millions of Google News articles learned gender stereotypes. Baur’s proposed solution is unbiased business and unbiased data. She should have read the academic paper; the authors suggest a specific way to treat bias.

— ProPublica releases the third installment in a four-part investigative series on black box algorithms that “dominate your life.”

Health and Medicine

— A team of investigators at the Medical University of South Carolina discovers that machine learning models that include biomarkers markedly improve clinicians’ ability to predict treatment outcomes.

— InsideBIGDATA touts a white paper on Healthcare and Life Sciences by Daniel Gutierrez.

DeepMind Rides the Subway

— Google DeepMind trains AI to navigate London’s Tube. PR blitz ensues. In MIT Technology Review, Will Knight describes the technology involved. Daily Mail reports, with pictures.

NVIDIA Supercomputers in Action

— OpenAI mines Reddit with a stupidly powerful NVIDIA DGX-1.

— NYU deploys a DGX-1 to support robotics research.

Marketplace

— Consulting firm ABI predicts a booming market for machine learning among telcos.

Applications

— Daven Mathies describes Netgear’s progress in facial recognition for home security.

— Can a computer win at poker? The answer reveals a lot about the direction of AI.

Autonomous Vehicles

— TuSimple, a Chinese AI startup, announces that it has broken ten records on the KITTI/Cityscapes public dataset for autonomous driving.

— Drive.ai seeks to develop tools to retrofit existing cars for autonomous operation.

Companies

— In-Q-Tel, the CIA’s venture capital arm, invests in Brainspace, a startup specializing in machine learning for unstructured data.