# Fully Distributed Actor-Critic Architecture for Multitask Deep Reinforcement Learning

arXiv: https://arxiv.org/abs/2110.12306 · HTML: https://mlsanity.com/paper/2110.12306/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 23 of its 71 references that are in the map.
- First posted on arXiv 23 Oct 2021.

## Builds on (most cited first, 23 of 23)

- [Adam: A Method for Stochastic Optimization](https://mlsanity.com/paper/1412.6980/) (2014)
- [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)
- [Convex optimization](https://mlsanity.com/paper/2106.01946/) (2021)
- [Asynchronous Methods for Deep Reinforcement Learning](https://mlsanity.com/paper/1602.01783/) (2016)
- [High-Dimensional Continuous Control Using Generalized Advantage Estimation](https://mlsanity.com/paper/1506.02438/) (2015)
- [OpenAI Gym](https://mlsanity.com/paper/1606.01540/) (2016)
- [IMPALA: Scalable Distributed Deep-RL with Importance Weighted Actor-Learner Architectures](https://mlsanity.com/paper/1802.01561/) (2018)
- [Distral: Robust Multitask Reinforcement Learning](https://mlsanity.com/paper/1707.04175/) (2017)
- [Modular Multitask Reinforcement Learning with Policy Sketches](https://mlsanity.com/paper/1611.01796/) (2016)
- [Fully Decentralized Multi-Agent Reinforcement Learning with Networked Agents](https://mlsanity.com/paper/1802.08757/) (2018)
- [Actor-Mimic: Deep Multitask and Transfer Reinforcement Learning](https://mlsanity.com/paper/1511.06342/) (2015)
- [Learning Continuous Control Policies by Stochastic Value Gradients](https://mlsanity.com/paper/1510.09142/) (2015)
- [Assessing Generalization in Deep Reinforcement Learning](https://mlsanity.com/paper/1810.12282/) (2018)
- [One-Shot Learning of Manipulation Skills with Online Dynamics Adaptation and Neural Network Priors](https://mlsanity.com/paper/1509.06841/) (2015)
- [Distributed Policy Evaluation Under Multiple Behavior Strategies](https://mlsanity.com/paper/1312.7606/) (2013)
- [Gossip-based Actor-Learner Architectures for Deep Reinforcement Learning](https://mlsanity.com/paper/1906.04585/) (2019)
- [Asynchronous Adaptation and Learning over Networks --- Part I: Modeling and Stability Analysis](https://mlsanity.com/paper/1312.5434/) (2013)
- [A Generalization of the Borkar-Meyn Theorem for Stochastic Recursive Inclusions](https://mlsanity.com/paper/1502.01953/) (2015)
- [Stochastic Recursive Inclusions in two timescales with non-additive iterate dependent Markov noise](https://mlsanity.com/paper/1611.05961/) (2016)
- [Diff-DAC: Distributed Actor-Critic for Average Multitask Deep Reinforcement Learning](https://mlsanity.com/paper/1710.10363/) (2017)
- [An Exact Distributed Newton Method for Reinforcement Learning](https://mlsanity.com/paper/1608.00070/) (2016)
- [Compatible features for Monotonic Policy Improvement](https://mlsanity.com/paper/1910.03880/) (2019)

## Built on by (most cited first, 1 of 1)

- [Policy Evaluation in Decentralized POMDPs with Belief Sharing](https://mlsanity.com/paper/2302.04151/) (2023)

## Abstract (arXiv)

We propose a fully distributed actor-critic architecture, named Diff-DAC, with application to multitask reinforcement learning (MRL). During the learning process, agents communicate their value and policy parameters to their neighbours, diffusing the information across a network of agents with no need for a central station. Each agent can only access data from its local task, but aims to learn a common policy that performs well for the whole set of tasks. The architecture is scalable, since the computational and communication cost per agent depends on the number of neighbours rather than the overall number of agents. We derive Diff-DAC from duality theory and provide novel insights into the actor-critic framework, showing that it is actually an instance of the dual ascent method. We prove almost sure convergence of Diff-DAC to a common policy under general assumptions that hold even for deep-neural network approximations. For more restrictive assumptions, we also prove that this common policy is a stationary point of an approximation of the original problem. Numerical results on multitask extensions of common continuous control benchmarks demonstrate that Diff-DAC stabilises learning and has a regularising effect that induces higher performance and better generalisation properties than previous architectures.