Research map / Multi-agent systems and games / Games and equilibria
Combo-Action: Training Agent For FPS Game with Auxiliary Tasks
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 Combo-Action with auxiliary tasks instead of primitive-action agents | FPS agent · ViZDoom death-match | Frags: 6.96 -> 27.86 | Helped |
| add Rotate Combo-Action (ablation) | Combo-Action agent with aiming ahead · ViZDoom ablation | Frags: 18.74 -> 26.35 | Helped |
| add aiming-ahead strategy (ablation) | Combo-Action agent with Rotate · ViZDoom ablation | Frags: 20.09 -> 26.35 | Helped |
| add aiming-ahead strategy (ablation) | Combo-Action agent without Rotate · ViZDoom ablation | Frags: 19.58 -> 18.74 | Hurt |
| add aiming-ahead strategy to random Combo-Action selection without Rotate (ablation) | random supervisory policy · ViZDoom ablation | Frags: 14.87 -> 13.44 | Hurt |
| learned supervisory network instead of random Combo-Action choice (ablation) | Combo-Action agent · ViZDoom ablation | Frags: 17.62 -> 26.35 | Helped |
| RPNmini detector instead of FasterRCNN | enemy detection · ViZDoom detection 50% overlap | miss rate: 22.71% -> 19.6% | Helped |
| RPNmini instead of FasterRCNN | enemy detection · ViZDoom detection | inference time: 0.61s -> 0.02s | Helped |
Closest papers by meaning
- Playing FPS Games with Deep Reinforcement Learning (AAAI 2017 · similarity 0.91)
- Training Agent for First-Person Shooter Game with Actor-Critic Curriculum Learning (ICLR 2017 · similarity 0.86)
- FightLadder: A Benchmark for Competitive Multi-Agent Reinforcement Learning (ICML 2024 · similarity 0.84)
- Advancing DRL Agents in Commercial Fighting Games: Training, Integration, and Agent-Human Alignment (ICML 2024 · similarity 0.84)
- Diffusion Models Are Real-Time Game Engines (ICLR 2025 · similarity 0.84)
- Deep Double Q-learning (NeurIPS 2026 · similarity 0.84)
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.