# Difference Rewards Policy Gradients

arXiv: https://arxiv.org/abs/2012.11258 · HTML: https://mlsanity.com/paper/2012.11258/

## Key facts

- Cited by 8 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 21 Dec 2020.

## Built on by (most cited first, 8 of 8)

- [Deep Multiagent Reinforcement Learning: Challenges and Directions](https://mlsanity.com/paper/2106.15691/) (2021)
- [More Centralized Training, Still Decentralized Execution: Multi-Agent Conditional Policy Factorization](https://mlsanity.com/paper/2209.12681/) (2022)
- [Factored Policy Gradients: Leveraging Structure for Efficient Learning in MOMDPs](https://mlsanity.com/paper/2102.10362/) (2021)
- [Deep Reinforcement Learning for Data-Driven Adaptive Scanning in Ptychography](https://mlsanity.com/paper/2203.15413/) (2022)
- [Zero-Shot Coordination in Ad Hoc Teams with Generalized Policy Improvement and Difference Rewards](https://mlsanity.com/paper/2510.16187/) (2025)
- [COSAC: Counterfactual Credit Assignment in Sequential Cooperative Teams](https://mlsanity.com/paper/2604.17693/) (2026)
- [Distributed Influence-Augmented Local Simulators for Parallel MARL in Large Networked Systems](https://mlsanity.com/paper/2207.00288/) (2022)
- [Marginal-Contribution Policy Gradients under Filtered Feedback for Multi-Agent LLMs](https://mlsanity.com/paper/2604.22785/) (2026)

## Abstract (arXiv)

Policy gradient methods have become one of the most popular classes of algorithms for multi-agent reinforcement learning. A key challenge, however, that is not addressed by many of these methods is multi-agent credit assignment: assessing an agent's contribution to the overall performance, which is crucial for learning good policies. We propose a novel algorithm called Dr.Reinforce that explicitly tackles this by combining difference rewards with policy gradients to allow for learning decentralized policies when the reward function is known. By differencing the reward function directly, Dr.Reinforce avoids difficulties associated with learning the Q-function as done by Counterfactual Multiagent Policy Gradients (COMA), a state-of-the-art difference rewards method. For applications where the reward function is unknown, we show the effectiveness of a version of Dr.Reinforce that learns an additional reward network that is used to estimate the difference rewards.