English

Neural Episodic Control

Machine Learning 2017-03-07 v1 Machine Learning

Abstract

Deep reinforcement learning methods attain super-human performance in a wide range of environments. Such methods are grossly inefficient, often taking orders of magnitudes more data than humans to achieve reasonable performance. We propose Neural Episodic Control: a deep reinforcement learning agent that is able to rapidly assimilate new experiences and act upon them. Our agent uses a semi-tabular representation of the value function: a buffer of past experience containing slowly changing state representations and rapidly updated estimates of the value function. We show across a wide range of environments that our agent learns significantly faster than other state-of-the-art, general purpose deep reinforcement learning agents.

Keywords

Cite

@article{arxiv.1703.01988,
  title  = {Neural Episodic Control},
  author = {Alexander Pritzel and Benigno Uria and Sriram Srinivasan and Adrià Puigdomènech and Oriol Vinyals and Demis Hassabis and Daan Wierstra and Charles Blundell},
  journal= {arXiv preprint arXiv:1703.01988},
  year   = {2017}
}
R2 v1 2026-06-22T18:37:24.136Z