Neural Population Learning beyond Symmetric Zero-sum Games
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2024
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866913191479476224 |
|---|---|
| author | Liu, Siqi Marris, Luke Lanctot, Marc Piliouras, Georgios Leibo, Joel Z. Heess, Nicolas |
| author_facet | Liu, Siqi Marris, Luke Lanctot, Marc Piliouras, Georgios Leibo, Joel Z. Heess, Nicolas |
| contents | We study computationally efficient methods for finding equilibria in n-player general-sum games, specifically ones that afford complex visuomotor skills. We show how existing methods would struggle in this setting, either computationally or in theory. We then introduce NeuPL-JPSRO, a neural population learning algorithm that benefits from transfer learning of skills and converges to a Coarse Correlated Equilibrium (CCE) of the game. We show empirical convergence in a suite of OpenSpiel games, validated rigorously by exact game solvers. We then deploy NeuPL-JPSRO to complex domains, where our approach enables adaptive coordination in a MuJoCo control domain and skill transfer in capture-the-flag. Our work shows that equilibrium convergent population learning can be implemented at scale and in generality, paving the way towards solving real-world games between heterogeneous players with mixed motives. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2401_05133 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Neural Population Learning beyond Symmetric Zero-sum Games Liu, Siqi Marris, Luke Lanctot, Marc Piliouras, Georgios Leibo, Joel Z. Heess, Nicolas Artificial Intelligence Multiagent Systems We study computationally efficient methods for finding equilibria in n-player general-sum games, specifically ones that afford complex visuomotor skills. We show how existing methods would struggle in this setting, either computationally or in theory. We then introduce NeuPL-JPSRO, a neural population learning algorithm that benefits from transfer learning of skills and converges to a Coarse Correlated Equilibrium (CCE) of the game. We show empirical convergence in a suite of OpenSpiel games, validated rigorously by exact game solvers. We then deploy NeuPL-JPSRO to complex domains, where our approach enables adaptive coordination in a MuJoCo control domain and skill transfer in capture-the-flag. Our work shows that equilibrium convergent population learning can be implemented at scale and in generality, paving the way towards solving real-world games between heterogeneous players with mixed motives. |
| title | Neural Population Learning beyond Symmetric Zero-sum Games |
| topic | Artificial Intelligence Multiagent Systems |
| url | https://arxiv.org/abs/2401.05133 |