Abstract

The deep reinforcement learning community has made several independent improvements to the DQN algorithm. However, it is unclear which of these extensions are complementary and can be fruitfully combined. This paper examines six extensions to the DQN algorithm and empirically studies their combination. Our experiments show that the combination provides state-of-the-art performance on the Atari 2600 benchmark, both in terms of data efficiency and final performance. We also provide results from a detailed ablation study that shows the contribution of each component to overall performance.

Keywords

Reinforcement learningBenchmark (surveying)Computer scienceArtificial intelligenceComponent (thermodynamics)ReinforcementMachine learningDeep learningEngineering

Affiliated Institutions

Related Publications

LightGCN

Xiangnan He , Kuan Deng , Xiang Wang +3 more

Graph Convolution Network (GCN) has become new state-of-the-art for collaborative filtering. Nevertheless, the reasons of its effectiveness for recommendation are not well under...

2020 3523 citations

Publication Info

Year
2018
Type
article
Volume
32
Issue
1
Citations
1630
Access
Closed

External Links

Download PDF (Free) View on DOI.org arXiv Semantic Scholar

Social Impact

Altmetric
PlumX Metrics

Social media, news, blog, policy document mentions

Citation Metrics

1630
OpenAlex
333
Influential
890
CrossRef

Cite This

Matteo Hessel, Joseph Modayil, Hado van Hasselt et al. (2018). Rainbow: Combining Improvements in Deep Reinforcement Learning. Proceedings of the AAAI Conference on Artificial Intelligence , 32 (1) . https://doi.org/10.1609/aaai.v32i1.11796

Identifiers

DOI
10.1609/aaai.v32i1.11796
arXiv
1710.02298

Data Quality

Data completeness: 84%