Reinforcement learning algorithms rely on carefully engineering environment rewards that are extrinsic to the agent. However, annotating each environment with hand-designed, dense rewards is not scalable, motivating the need for developing reward functions that are intrinsic to the agent. Curiosity is a type of intrinsic reward function which uses prediction error as reward signal.

In this paper:
(a) We perform the first large-scale study of purely curiosity-driven learning, i.e. without any extrinsic rewards, across 54 standard benchmark environments, including the Atari game suite. Our results show surprisingly good performance, and a high degree of alignment between the intrinsic curiosity objective and the hand-designed extrinsic rewards of many game environments.
(b) We investigate the effect of using different feature spaces for computing prediction error and show that random features are sufficient for many popular RL game benchmarks, but learned features appear to generalize better (e.g. to novel game levels in Super Mario Bros.).
(c) We demonstrate limitations of the prediction-based rewards in stochastic setups.

A snapshot of the 54 environments investigated in the paper. We show that agents are able to makeprogress using no extrinsic reward, or end-of-episode signal, and only using curiosity.

We have released the TensorFlow based implementation on the github page. Try our code!

@inproceedings{pathak18largescale,
  Author = {Burda, Yuri and
  Edwards, Harri and Pathak, Deepak and
  Storkey, Amos and Darrell, Trevor and
  Efros, Alexei A.},
  Title = {Large-Scale Study of
  Curiosity-Driven Learning},
  Booktitle = {ICLR},
  Year = {2019}
}