Model based Multi-agent Reinforcement Learning with Tensor Decompositions
arXiv:2110.14524
Abstract
A challenge in multi-agent reinforcement learning is to be able to generalize over intractable state-action spaces. Inspired from Tesseract [Mahajan et al., 2021], this position paper investigates generalisation in state-action space over unexplored state-action pairs by modelling the transition and reward functions as tensors of low CP-rank. Initial experiments on synthetic MDPs show that using tensor decompositions in a model-based reinforcement learning algorithm can lead to much faster convergence if the true transition and reward functions are indeed of low rank.
References in corpus (8)
- A Survey of Model Compression and Acceleration for Deep Neural Networks
- QPLEX: Duplex Dueling Multi-Agent Q-Learning
- Provable Tensor Factorization with Missing Data
- MAVEN: Multi-Agent Variational Exploration
- RODE: Learning Roles to Decompose Multi-Agent Tasks
- Open-Ended Learning Leads to Generally Capable Agents
- UneVEn: Universal Value Exploration for Multi-Agent Reinforcement Learning
- Tesseract: Tensorised Actors for Multi-Agent Reinforcement Learning