Ken Stanley is also moving from Uber AI Labs to OpenAI

Just like Jeff Clune a few months ago:

twitter.com/kenneth0stanley/status/1253347502897668097

"where I will be starting a group (and indeed hiring) focused on achieving open-endedness in machine learning."

(I should have started using Twitter more actively long ago; that's how one gets quite a bit of info these days.)

AI Feynman: a Physics-Inspired Method for Symbolic Regression

By Silviu-Marian Udrescu and Max Tegmark: arxiv.org/abs/1905.11481

github.com/SJ001/AI-Feynman

"recursive multidimensional symbolic regression algorithm that combines neural network fitting with a suite of physics-inspired techniques."

Neural networks are used to discover "symmetries, separability, compositionality and other simplifying properties" and thus reduce the search space.

"We apply it to 100 equations from the Feynman Lectures on Physics, and it discovers all of them, while previous publicly available software cracks only 71; for a more difficult test set, we improve the state of the art success rate from 15% to 90%."

Follow-up by the same authors: "Symbolic Pregression: Discovering Physical Laws from Raw Distorted Video": arxiv.org/abs/2005.11212