Exploration by Random Network Distillation

10/30/2018
by   Yuri Burda, et al.
0

We introduce an exploration bonus for deep reinforcement learning methods that is easy to implement and adds minimal overhead to the computation performed. The bonus is the error of a neural network predicting features of the observations given by a fixed randomly initialized neural network. We also introduce a method to flexibly combine intrinsic and extrinsic rewards. We find that the random network distillation (RND) bonus combined with this increased flexibility enables significant progress on several hard exploration Atari games. In particular we establish state of the art performance on Montezuma's Revenge, a game famously difficult for deep reinforcement learning methods. To the best of our knowledge, this is the first method that achieves better than average human performance on this game without using demonstrations or having access to the underlying state of the game, and occasionally completes the first level.

READ FULL TEXT
research
02/06/2023

Intrinsic Rewards from Self-Organizing Feature Maps for Exploration in Reinforcement Learning

We introduce an exploration bonus for deep reinforcement learning method...
research
05/18/2019

Combining Experience Replay with Exploration by Random Network Distillation

Our work is a simple extension of the paper "Exploration by Random Netwo...
research
09/22/2021

On Bonus-Based Exploration Methods in the Arcade Learning Environment

Research on exploration in reinforcement learning, as applied to Atari 2...
research
10/05/2022

On Neural Consolidation for Transfer in Reinforcement Learning

Although transfer learning is considered to be a milestone in deep reinf...
research
09/05/2021

An Exploration of Deep Learning Methods in Hungry Geese

Hungry Geese is a n-player variation of the popular game snake. This pap...
research
12/31/2021

SimSR: Simple Distance-based State Representation for Deep Reinforcement Learning

This work explores how to learn robust and generalizable state represent...

Please sign up or login with your details

Forgot password? Click here to reset