Coordinated Multi-Agent Imitation Learning
arXiv:1703.03121
Abstract
We study the problem of imitation learning from demonstrations of multiple coordinating agents. One key challenge in this setting is that learning a good model of coordination can be difficult, since coordination is often implicit in the demonstrations and must be inferred as a latent variable. We propose a joint approach that simultaneously learns a latent coordination model along with the individual policies. In particular, our method integrates unsupervised structure learning with conventional imitation learning. We illustrate the power of our approach on a difficult problem of learning multiple policies for fine-grained behavior modeling in team sports, where different players occupy different roles in the coordinated team strategy. We show that having a coordination model to infer the roles of players yields substantially improved imitation loss compared to conventional baselines.
arXiv commentInternational Conference on Machine Learning 2017
References in corpus (3)
Cited by in corpus (26)
- Learning Attentional Communication for Multi-Agent Cooperation
- State Alignment-based Imitation Learning
- Real World Games Look Like Spinning Tops
- baller2vec: A Multi-Entity Transformer For Multi-Agent Spatiotemporal Modeling
- Inverse Dynamic Games Based on Maximum Entropy Inverse Reinforcement Learning
- Learning Hierarchical Teaching Policies for Cooperative Agents
- Learning Self-Correctable Policies and Value Functions from Demonstrations with Negative Sampling
- Parallel Knowledge Transfer in Multi-Agent Reinforcement Learning
- Scalable Multi-Agent Inverse Reinforcement Learning via Actor-Attention-Critic
- A Q-values Sharing Framework for Multiagent Reinforcement Learning under Budget Constraint
- MARL with General Utilities via Decentralized Shadow Reward Actor-Critic
- A Graph Attention Based Approach for Trajectory Prediction in Multi-agent Sports Games
- Skill Discovery of Coordination in Multi-agent Reinforcement Learning
- Modeling and Interpreting Real-world Human Risk Decision Making with Inverse Reinforcement Learning
- Multi-Agent Interactions Modeling with Correlated Policies
- DeepHoops: Evaluating Micro-Actions in Basketball Using Deep Feature Representations of Spatio-Temporal Data
- Crowd Density Forecasting by Modeling Patch-based Dynamics
- Improved Structural Discovery and Representation Learning of Multi-Agent Data
- Learning interaction rules from multi-animal trajectories via augmented behavioral models
- Time-series Imputation of Temporally-occluded Multiagent Trajectories
- Predicting the outcome of team movements -- Player time series analysis using fuzzy and deep methods for representation learning
- Decentralized Multi-Agents by Imitation of a Centralized Controller
- Learning Calibratable Policies using Programmatic Style-Consistency
- Dynamic Relational Inference in Multi-Agent Trajectories
- Multi-Agent Imitation Learning with Copulas
- next-gen-scraPy: Extracting NFL Tracking Data from Images to Evaluate Quarterbacks and Pass Defenses