ADDQ: Adaptive distributional double Q-learning

Leif Döring, Benedikt Wille, Maximilian Birr, Mihail Bı̂rsan, Martin Slowik
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:14344-14390, 2025.

Abstract

Bias problems in the estimation of Q-values are a well-known obstacle that slows down convergence of Q-learning and actor-critic methods. One of the reasons of the success of modern RL algorithms is partially a direct or indirect overestimation reduction mechanism. We introduce an easy to implement method built on top of distributional reinforcement learning (DRL) algorithms to deal with the overestimation in a locally adaptive way. Our framework ADDQ is simple to implement, existing DRL implementations can be improved with a few lines of code. We provide theoretical backup and experimental results in tabular, Atari, and MuJoCo environments, comparisons with state-of-the-art methods, and a proof of convergence in the tabular case.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-doring25a, title = {{ADDQ}: Adaptive distributional double Q-learning}, author = {D\"{o}ring, Leif and Wille, Benedikt and Birr, Maximilian and B\^{\i}rsan, Mihail and Slowik, Martin}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {14344--14390}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/doring25a/doring25a.pdf}, url = {https://proceedings.mlr.press/v267/doring25a.html}, abstract = {Bias problems in the estimation of Q-values are a well-known obstacle that slows down convergence of Q-learning and actor-critic methods. One of the reasons of the success of modern RL algorithms is partially a direct or indirect overestimation reduction mechanism. We introduce an easy to implement method built on top of distributional reinforcement learning (DRL) algorithms to deal with the overestimation in a locally adaptive way. Our framework ADDQ is simple to implement, existing DRL implementations can be improved with a few lines of code. We provide theoretical backup and experimental results in tabular, Atari, and MuJoCo environments, comparisons with state-of-the-art methods, and a proof of convergence in the tabular case.} }
Endnote
%0 Conference Paper %T ADDQ: Adaptive distributional double Q-learning %A Leif Döring %A Benedikt Wille %A Maximilian Birr %A Mihail Bı̂rsan %A Martin Slowik %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-doring25a %I PMLR %P 14344--14390 %U https://proceedings.mlr.press/v267/doring25a.html %V 267 %X Bias problems in the estimation of Q-values are a well-known obstacle that slows down convergence of Q-learning and actor-critic methods. One of the reasons of the success of modern RL algorithms is partially a direct or indirect overestimation reduction mechanism. We introduce an easy to implement method built on top of distributional reinforcement learning (DRL) algorithms to deal with the overestimation in a locally adaptive way. Our framework ADDQ is simple to implement, existing DRL implementations can be improved with a few lines of code. We provide theoretical backup and experimental results in tabular, Atari, and MuJoCo environments, comparisons with state-of-the-art methods, and a proof of convergence in the tabular case.
APA
Döring, L., Wille, B., Birr, M., Bı̂rsan, M. & Slowik, M.. (2025). ADDQ: Adaptive distributional double Q-learning. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:14344-14390 Available from https://proceedings.mlr.press/v267/doring25a.html.

Related Material