Towards General and Autonomous Learning of Core Skills: A Case Study in Locomotion | DeepMind

Towards General and Autonomous Learning of Core Skills: A Case Study in Locomotion | DeepMind

Modern Reinforcement Learning (RL) algorithms promise to solve difficult motor control problems directly from raw sensory inputs. Their attraction is due in part to the fact that they can represent a general class of methods that allow to learn a solution with a reasonably set reward and minimal prior knowledge, even in situations where it is difficult or expensive for a human expert.

2 mentions: @DeepMind@ozanozdil
Date: 2020/09/30 09:52

Referring Tweets

@DeepMind Learning to walk with 2+ legs via general purpose RL: researchers show how a single agent learns versatile locomotion skills for several legged robots using only on-board sensors for state estimation & reward. Blog: t.co/SEq0y3TO7b Paper: t.co/fx4yoJmJAj

Related Entries

Read more Course Readings | The Human Intelligence Enterprise | Electrical Engineering and Computer Science | ...
0 users, 1 mentions 2020/05/21 14:21
Read more VQA: Visual Question Answering
0 users, 2 mentions 2020/06/12 15:51
Read more [1910.00553v2] Better Document-Level Machine Translation with Bayes' Ruleopen searchopen navigation ...
0 users, 1 mentions 2020/07/06 17:21
Read more GitHub - deepmind/optax: Optax is a gradient processing and optimization library for JAX.
0 users, 2 mentions 2020/08/17 12:53
Read more Understanding meta-trained algorithms through a Bayesian lens | by DeepMind Safety Research | Dec, 2...
0 users, 1 mentions 2020/12/03 16:03