Improved Communication Efficiency in Federated Natural Policy Gradient via ADMM-based Gradient Updates.
Guangchen LanHan WangJames AndersonChristopher G. BrintonVaneet AggarwalPublished in: NeurIPS (2023)
Keyphrases
- policy gradient
- gradient method
- reinforcement learning
- parametric optimization
- function approximation
- actor critic
- convex optimization
- reinforcement learning algorithms
- model free reinforcement learning
- neural network
- optimal control
- approximation methods
- machine learning
- multi agent
- computational complexity
- variance reduction