On the Stability of Learning in Network Games with Many Players
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_ | 1866910379814158336 |
|---|---|
| author | Hussain, Aamal Leonte, Dan Belardinelli, Francesco Piliouras, Georgios |
| author_facet | Hussain, Aamal Leonte, Dan Belardinelli, Francesco Piliouras, Georgios |
| contents | Multi-agent learning algorithms have been shown to display complex, unstable behaviours in a wide array of games. In fact, previous works indicate that convergent behaviours are less likely to occur as the total number of agents increases. This seemingly prohibits convergence to stable strategies, such as Nash Equilibria, in games with many players.
To make progress towards addressing this challenge we study the Q-Learning Dynamics, a classical model for exploration and exploitation in multi-agent learning. In particular, we study the behaviour of Q-Learning on games where interactions between agents are constrained by a network. We determine a number of sufficient conditions, depending on the game and network structure, which guarantee that agent strategies converge to a unique stable strategy, called the Quantal Response Equilibrium (QRE). Crucially, these sufficient conditions are independent of the total number of agents, allowing for provable convergence in arbitrarily large games.
Next, we compare the learned QRE to the underlying NE of the game, by showing that any QRE is an $ε$-approximate Nash Equilibrium. We first provide tight bounds on $ε$ and show how these bounds lead naturally to a centralised scheme for choosing exploration rates, which enables independent learners to learn stable approximate Nash Equilibrium strategies. We validate the method through experiments and demonstrate its effectiveness even in the presence of numerous agents and actions. Through these results, we show that independent learning dynamics may converge to approximate Nash Equilibria, even in the presence of many agents. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2403_15848 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | On the Stability of Learning in Network Games with Many Players Hussain, Aamal Leonte, Dan Belardinelli, Francesco Piliouras, Georgios Computer Science and Game Theory 93A16, 91A26, 91A68, 58K35 G.3; J.4; F.2.2 Multi-agent learning algorithms have been shown to display complex, unstable behaviours in a wide array of games. In fact, previous works indicate that convergent behaviours are less likely to occur as the total number of agents increases. This seemingly prohibits convergence to stable strategies, such as Nash Equilibria, in games with many players. To make progress towards addressing this challenge we study the Q-Learning Dynamics, a classical model for exploration and exploitation in multi-agent learning. In particular, we study the behaviour of Q-Learning on games where interactions between agents are constrained by a network. We determine a number of sufficient conditions, depending on the game and network structure, which guarantee that agent strategies converge to a unique stable strategy, called the Quantal Response Equilibrium (QRE). Crucially, these sufficient conditions are independent of the total number of agents, allowing for provable convergence in arbitrarily large games. Next, we compare the learned QRE to the underlying NE of the game, by showing that any QRE is an $ε$-approximate Nash Equilibrium. We first provide tight bounds on $ε$ and show how these bounds lead naturally to a centralised scheme for choosing exploration rates, which enables independent learners to learn stable approximate Nash Equilibrium strategies. We validate the method through experiments and demonstrate its effectiveness even in the presence of numerous agents and actions. Through these results, we show that independent learning dynamics may converge to approximate Nash Equilibria, even in the presence of many agents. |
| title | On the Stability of Learning in Network Games with Many Players |
| topic | Computer Science and Game Theory 93A16, 91A26, 91A68, 58K35 G.3; J.4; F.2.2 |
| url | https://arxiv.org/abs/2403.15848 |