Bits & Bytes

  • Interesting article in Science about explainability approaches for deep neural networks. Marco Ribeiro and Carlos Guestrin’s LIME is discussed—check out my interview with Carlos for more detail on that project. This recent article on interpreting neurons in an LSTM network is related, and also very interesting.
  • Impressive work by the EFF compiling a bunch of metrics of progress in ML/AI. Nice touch presenting it as a Jupyter notebook!
  • Researchers from Stanford and iRythm collaborate to develop 34-layer CNN that can detect arrhythmias in single-lead ECG better than a cardiologist.
  • Three new papers from Google DeepMind explore teaching complex movement to simulated humanoid forms.
  • Baidu releases Apollo—a comprehensive open source platform for self-driving cars. Is this the future Android OS for the autonomous, connected vehicle? They announced with a huge partner ecosystem, so it will be interesting to see where this goes.

Sign up for our Newsletter to receive the Bits & Bytes weekly to your inbox.