Research map / Multi-agent systems and games / Games and equilibria
AIVAT: A New Variance Reduction Technique for Agent Evaluation in Imperfect Information Games
Neil Burch, Martin Schmid, Matej Moravcik, Dustin Morill, Michael Bowling
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 AIVAT with both players' strategies known instead of raw chip counts | Leduc hold'em self-play · Leduc hold'em self-play | standard deviation: 99.9% | Improved |
| Use AIVAT with one player's strategy known instead of raw chip counts | Leduc hold'em self-play · Leduc hold'em self-play | standard deviation: 99.8% | Improved |
| Use MIVAT instead of raw chip counts | Leduc hold'em self-play · Leduc hold'em self-play | standard deviation: 33.8% | Improved |
| Use MIVAT+IO_x instead of raw chip counts | Leduc hold'em self-play · Leduc hold'em self-play | standard deviation: 45.1% | Improved |
| Use AIVAT instead of basic chip-count estimates | Leduc hold'em with dissimilar strategies · Leduc hold'em, dissimilar strategies | standard deviation: 48% to 75% | Improved |
| Use MIVAT plus imaginary observations instead of basic chip-count estimates | Leduc hold'em with dissimilar strategies · Leduc hold'em, dissimilar strategies | standard deviation: 25% | Improved |
| Use AIVAT instead of the unmodified game value | HUNL · HUNL self-play and dissimilar strategies | standard deviation: a bit more than 68% | Improved |
| Use MIVAT with imaginary observations instead of the unmodified game value | HUNL · HUNL self-play and dissimilar strategies | standard deviation: 39% | Improved |
| Use MIVAT instead of the unmodified game value | HUNL · HUNL | standard deviation: 18% | Improved |
Closest papers by meaning
- Multiagent Evaluation under Incomplete Information (NeurIPS 2019 · similarity 0.87)
- Outbidding and Outbluffing Elite Humans: Mastering Liar’s Poker via Self-Play and Reinforcement Learning (NeurIPS 2026 · similarity 0.86)
- Depth-Limited Solving for Imperfect-Information Games (NeurIPS 2018 · similarity 0.86)
- Preference-CFR: Beyond Nash Equilibrium for Better Game Strategies (ICML 2025 · similarity 0.86)
- PokerBench: Training Large Language Models to Become Professional Poker Players (AAAI 2025 · similarity 0.84)
- Safe and Robust Subgame Exploitation in Imperfect Information Games (ICML 2024 · 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.
