Research map / Multi-agent systems and games / Human-AI and goal-conditioned
HogRider: Champion Agent of Microsoft Malmo Collaborative AI Challenge
Yanhai Xiong, Haipeng Chen, Mengchen Zhao, Bo An
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use Separate instead of Weighted action selection (ablation) | HogRider · MCAC per step | mean score: 2.527 -> 2.752 | Improved |
| use Separate instead of Mixed action selection (ablation) | HogRider · MCAC per step | mean score: 2.478 -> 2.752 | Improved |
| use Separate instead of Mixed action selection (ablation) | HogRider · MCAC per step | variance/mean: 1.1340 -> 0.9266 | Improved |
| use G-Bayes instead of Bayes for type belief update (ablation) | HogRider · MCAC per step | overall detection accuracy: 0.672 -> 0.747 | Improved |
| add tanh squashing to G-Bayes belief update (ablation) | HogRider · MCAC per step | mean score: 2.505 -> 2.752 | Improved |
| add tanh squashing to G-Bayes belief update (ablation) | HogRider · MCAC per step | variance/mean: 0.9062 -> 0.9266 | Hurt |
| add tanh squashing to G-Bayes belief update (ablation) | HogRider · MCAC per step | overall detection accuracy: 0.747 -> 0.937 | Improved |
| use G-Bayes+tanh instead of traditional Bayes (ablation) | HogRider · MCAC per step | mean score: 2.456 -> 2.752 | Improved |
| add human reasoning-derived warm start of Q-function (ablation) | tabular Q-learning · Pig Chase vs Focused agent, validation | mean score: 10.9% improvement | Improved |
| add human reasoning-derived warm start of Q-function (ablation) | tabular Q-learning · Pig Chase vs Focused agent, validation | variance/mean: 5.7% decrease | Improved |
| use active epsilon-greedy instead of Deterministic exploration (ablation) | tabular Q-learning · Pig Chase vs Focused agent, validation | mean score: 4.1% edge | Improved |
| use active epsilon-greedy instead of epsilon-greedy exploration (ablation) | tabular Q-learning · Pig Chase vs Focused agent, validation | mean score: 6.7% edge | Improved |
Closest papers by meaning
- FightLadder: A Benchmark for Competitive Multi-Agent Reinforcement Learning (ICML 2024 · similarity 0.86)
- MARSHAL: Incentivizing Multi-Agent Reasoning via Self-Play with Strategic LLMs (ICLR 2026 · similarity 0.86)
- Sparks of Cooperative Reasoning: LLMs as Strategic Hanabi Agents (ICML 2026 · similarity 0.85)
- Mimicking To Dominate: Imitation Learning Strategies for Success in Multiagent Games (NeurIPS 2024 · similarity 0.85)
- Training Generalizable Collaborative Agents via Strategic Risk Aversion (NeurIPS 2026 · similarity 0.85)
- AgentCDM: Enhancing Multi-Agent Collaborative Decision-Making via ACH-Inspired Structured Reasoning (AAAI 2026 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
