Really cool work by @pathak2206 and Chris Lu. Multiple agents learn to self-assemble and perform locomotion tasks! Done in Unity w/ Ml-Agents. https://t.co/E3luD5yTNr https://t.co/a1fnlVbh1l
I really like this work! One of the more impressive models for learning the dynamics of complex environments (even from pixels). It seems to open up a lot of possibilities for model-based RL.
Today we are excited to release video recordings of lectures from "Advanced Deep Learning and Reinforcement Learning", a course on deep RL taught at @UCL earlier this year by DeepMind researchers:
https://t.co/znsWtTxQcN
Enjoy!
Learning to play Atari games without rewards by combining human preferences and expert demonstrations (collaboration with @OpenAI): https://t.co/tcdVFCynjf
Need a tool that automatically searches over neural architectures, learning to combine the best ones into a high-quality model? Check out AdaNet, a lightweight @TensorFlow-based framework for automatically learning models with minimal expert intervention. https://t.co/s7K0sYOdS3