# Revisiting Action Factorization for Complex Action Spaces

arXiv: https://arxiv.org/abs/2606.26574 · HTML: https://mlsanity.com/paper/2606.26574/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 13 of its 46 references that are in the map.
- First posted on arXiv 25 Jun 2026.

## Builds on (most cited first, 13 of 13)

- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [Continuous control with deep reinforcement learning](https://mlsanity.com/paper/1509.02971/) (2015)
- [Playing Atari with Deep Reinforcement Learning](https://mlsanity.com/paper/1312.5602/) (2013)
- [Categorical Reparameterization with Gumbel-Softmax](https://mlsanity.com/paper/1611.01144/) (2016)
- [DeepMind Control Suite](https://mlsanity.com/paper/1801.00690/) (2018)
- [Value-Decomposition Networks For Cooperative Multi-Agent Learning](https://mlsanity.com/paper/1706.05296/) (2017)
- [QPLEX: Duplex Dueling Multi-Agent Q-Learning](https://mlsanity.com/paper/2008.01062/) (2020)
- [Parametrized Deep Q-Networks Learning: Reinforcement Learning with Discrete-Continuous Hybrid Action Space](https://mlsanity.com/paper/1810.06394/) (2018)
- [Hybrid Actor-Critic Reinforcement Learning in Parameterized Action Space](https://mlsanity.com/paper/1903.01344/) (2019)
- [Gymnasium: A Standard Interface for Reinforcement Learning Environments](https://mlsanity.com/paper/2407.17032/) (2024)
- [Discrete and Continuous Action Representation for Practical RL in Video Games](https://mlsanity.com/paper/1912.11077/) (2019)
- [HyAR: Addressing Discrete-Continuous Action Reinforcement Learning via Hybrid Action Representation](https://mlsanity.com/paper/2109.05490/) (2021)
- [Autoregressive Policies for Continuous Control Deep Reinforcement Learning](https://mlsanity.com/paper/1903.11524/) (2019)

## Abstract (arXiv)

Many real-world control problems involve hybrid discrete-continuous action spaces. For example, steering and signaling in autonomous driving, and aiming and firing in robotics or video-games. Despite real-world hybrid factorization and reinforcement learning framework support for complex action spaces (e.g., Gymnasium, PettingZoo, TorchRL, SeedRL, Mujoco, etc), the default environments within those frameworks often implement uniform action space configurations (LunarLander, Walker2D, Cheetah, SMAC, SUMO, Ant, Atari). Landmark hybrid-action benchmarks (RoboCup 2D HFO, SC2LE, Platform, CARLA, etc) are mostly heavyweight or archival implementations originating from papers which test one or a small number of competing factorization methods on one kind of control. This article provides a cross-sectional study of factorization methods [independent networks, shared encoder, VDN, QPLEX, Joint, Auto-Regressive] on each of three families of algorithms [PPO, SAC, DQN] across three action spaces [discretized, hybrid, continuous] over four lightweight environments [Platform, hybrid-LunarLander, Hybrid-Shoot, CoopPush]. Accounting for some invalid pairings such as joint-continuous, we are left with 220 configurations to analyze each method. We provide two new C++ parallel gymnasium and petting-zoo compliant environments [CoopPush, Hybrid-Shoot] to isolate particular challenges such as state-dependent inter-action dependence. Finally, we introduce VDN-PPO and PPO-MIX which use a branching critic to assign credit to multi-headed PPO. These variants out-perform all other tested PPO factorizations. Our results suggest that branching dueling architectures balance compute and performance most effectively, with Auto-Regressive actions reaching the highest performance overall and native continuous SAC outperforming discrete and hybrid algorithms, albiet both at increased computational cost.