Need autonomous driving training data? ›

What Happened in AI This Month: February 2017 Edition

What Happened in AI This Month: February 2017 Edition

New blog series! Would you like a monthly rundown on AI news, talks, acquisitions, etc.? We knew you would so we’re going to start doing it!

Here’s what happened in February. Enjoy.

  • WIRED dove into how Libratus, the system that beat some of the best players in the world at poker, works. Read.
  • The California DMV’s annual “disengagement reports” on self-driving cars showed significant and quick progress in autonomous vehicles. Read.
  • Quartz live-streamed their bot’s progress in learning Atari games. Watch and read.
  • The Atlantic explored what autonomous AI would mean for Christianity. Read.
  • Bloomberg covered Sentient Technologies’ rad AI hedge fund. Read.
  • Pew Research Center and Elon University’s Imagining the Internet Center canvassed over a thousand experts on the pros and cons of the spread of algorithms. Read.
  • Ford announced plans to invest $1 billion over 5 years into Argo AI, a stealthy startup founded by Google and Uber vets. Read.
  • Some machine learning experts made an excellent list of useful public datasets for AI. Read.
  • Um. Graphcore made images/maps/graphs of phases of the machine learning process—and they look like brain scans and it’s nuts. See.
  • The R&D team at SVDS did a review and ranking of available deep learning tools. Read.
  • Google Cloud Platform announced a Kaggle competition to develop classification algorithms from the YouTube-8M V2 dataset that produce video-level labels. Read.
  • OpenAI explained adversarial examples in machine learning and why they’re tough to protect against. Read.
  • GM said it’s partnering with Lyft to deploy test fleets of thousands of self-driving electric cars next year. Read.
  • Ed Newton-Rex, founder of Jukedeck, compiled a list of all the things AIs can do today. Read.
  • Stephen Pratt, Noodle.ai founder and CEO, gave a talk at Strata + Hadoop World on figuring out and maximizing AI’s value in the enterprise. Watch/listen.
  • Researchers proposed a more complex, more human-like architecture and training method for deep neural networks. Read.

image credit: Yanina via CC0 1.0