Centralized Training for Decentralized Execution, where agents are trained offline in a centralized fashion and execute online in a decentralized manner, has become a popular approach in Multi-Agent Reinforcement Learning (MARL). In particular, it has become popular to develop actor-critic methods that train decentralized actors with a centralized critic where the centralized critic is allowed access to global information of the entire system, including the true system state. Such centralized critics are possible given offline information and are not used for online execution. While these methods perform well in a number of domains and have become a de facto standard in MARL, using a centralized critic in this context has yet to be sufficiently analyzed theoretically or empirically. In this paper, we therefore formally analyze centralized and decentralized critic approaches, and analyze the effect of using state-based critics in partially observable environments. We derive theories contrary to the common intuition: critic centralization is not strictly beneficial, and using state values can be harmful. We further prove that, in particular, state-based critics can introduce unexpected bias and variance compared to history-based critics. Finally, we demonstrate how the theory applies in practice by comparing different forms of critics on a wide range of common multi-agent benchmarks. The experiments show practical issues such as the difficulty of representation learning with partial observability, which highlights why the theoretical problems are often overlooked in the literature.
more »
« less
A Deeper Understanding of State-Based Critics in Multi-Agent Reinforcement Learning
Centralized Training for Decentralized Execution, where training is done in a centralized offline fashion, has become a popular solution paradigm in Multi-Agent Reinforcement Learning. Many such methods take the form of actor-critic with state-based critics, since centralized training allows access to the true system state, which can be useful during training despite not being available at execution time. State-based critics have become a common empirical choice, albeit one which has had limited theoretical justification or analysis. In this paper, we show that state-based critics can introduce bias in the policy gradient estimates, potentially undermining the asymptotic guarantees of the algorithm. We also show that, even if the state-based critics do not introduce any bias, they can still result in a larger gradient variance, contrary to the common intuition. Finally, we show the effects of the theories in practice by comparing different forms of centralized critics on a wide range of common benchmarks, and detail how various environmental properties are related to the effectiveness of different types of critics.
more »
« less
- NSF-PAR ID:
- 10315765
- Date Published:
- Journal Name:
- Proceedings of the AAAI Conference on Artificial Intelligence
- ISSN:
- 2374-3468
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
In partially observable reinforcement learning, offline training gives access to latent information which is not available during online training and/or execution, such as the system state. Asymmetric actor-critic methods exploit such information by training a history-based policy via a state-based critic. However, many asymmetric methods lack theoretical foundation, and are only evaluated on limited domains. We examine the theory of asymmetric actor-critic methods which use state-based critics, and expose fundamental issues which undermine the validity of a common variant, and limit its ability to address partial observability. We propose an unbiased asymmetric actor-critic variant which is able to exploit state information while remaining theoretically sound, maintaining the validity of the policy gradient theorem, and introducing no bias and relatively low variance into the training process. An empirical evaluation performed on domains which exhibit significant partial observability confirms our analysis, demonstrating that unbiased asymmetric actor-critic converges to better policies and/or faster than symmetric and biased asymmetric baselines.more » « less
-
Centralized Training for Decentralized Execution, where agents are trained offline using centralized information but execute in a decentralized manner online, has gained popularity in the multi-agent reinforcement learning community. In particular, actor-critic methods with a centralized critic and decentralized actors are a common instance of this idea. However, the implications of using a centralized critic in this context are not fully discussed and understood even though it is the standard choice of many algorithms. We therefore formally analyze centralized and decentralized critic approaches, providing a deeper understanding of the implications of critic choice. Because our theory makes unrealistic assumptions, we also empirically compare the centralized and decentralized critic methods over a wide set of environments to validate our theories and to provide practical advice. We show that there exist misconceptions regarding centralized critics in the current literature and show that the centralized critic design is not strictly beneficial, but rather both centralized and decentralized critics have different pros and cons that should be taken into account by algorithm designersmore » « less
-
We consider the problem of spectrum sharing by multiple cellular operators. We propose a novel deep Reinforcement Learning (DRL)-based distributed power allocation scheme which utilizes the multi-agent Deep Deterministic Policy Gradient (MA-DDPG) algorithm. In particular, we model the base stations (BSs) that belong to the multiple operators sharing the same band, as DRL agents that simultaneously determine the transmit powers to their scheduled user equipment (UE) in a synchronized manner. The power decision of each BS is based on its own observation of the radio environment (RF) environment, which consists of interference measurements reported from the UEs it serves, and a limited amount of information obtained from other BSs. One advantage of the proposed scheme is that it addresses the single-agent non-stationarity problem of RL in the multi-agent scenario by incorporating the actions and observations of other BSs into each BS's own critic which helps it to gain a more accurate perception of the overall RF environment. A centralized-training-distributed-execution framework is used to train the policies where the critics are trained over the joint actions and observations of all BSs while the actor of each BS only takes the local observation as input in order to produce the transmit power. Simulation with the 6 GHz Unlicensed National Information Infrastructure (U-NII)-5 band shows that the proposed power allocation scheme can achieve better throughput performance than several state-of-the-art approaches.more » « less
-
Providing privacy protection has been one of the primary motivations of Federated Learning (FL). Recently, there has been a line of work on incorporating the formal privacy notion of differential privacy with FL. To guarantee the client-level differential privacy in FL algorithms, the clients’ transmitted model updates have to be clipped before adding privacy noise. Such clipping operation is substantially different from its counterpart of gradient clipping in the centralized differentially private SGD and has not been well-understood. In this paper, we first empirically demonstrate that the clipped FedAvg can perform surprisingly well even with substantial data heterogeneity when training neural networks, which is partly because the clients’ updates become similar for several popular deep architectures. Based on this key observation, we provide the convergence analysis of a differential private (DP) FedAvg algorithm and highlight the relationship between clipping bias and the distribution of the clients’ updates. To the best of our knowledge, this is the first work that rigorously investigates theoretical and empirical issues regarding the clipping operation in FL algorithms.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- The DOI is not currently available.
