Penalty-regulated dynamics and robust learning procedures in games
From MaRDI portal
Publication:3449451
DOI10.1287/MOOR.2014.0687zbMATH Open1377.91033OpenAlexW2116718598WikidataQ60142071 ScholiaQ60142071MaRDI QIDQ3449451FDOQ3449451
Authors: Pierre Coucheney, Bruno Gaujal, Panayotis Mertikopoulos 
Publication date: 4 November 2015
Published in: Mathematics of Operations Research (Search for Journal in Brave)
Abstract: Starting from a heuristic learning scheme for N-person games, we derive a new class of continuous-time learning dynamics consisting of a replicator-like drift adjusted by a penalty term that renders the boundary of the game's strategy space repelling. These penalty-regulated dynamics are equivalent to players keeping an exponentially discounted aggregate of their on-going payoffs and then using a smooth best response to pick an action based on these performance scores. Owing to this inherent duality, the proposed dynamics satisfy a variant of the folk theorem of evolutionary game theory and they converge to (arbitrarily precise) approximations of Nash equilibria in potential games. Motivated by applications to traffic engineering, we exploit this duality further to design a discrete-time, payoff-based learning algorithm which retains these convergence properties and only requires players to observe their in-game payoffs: moreover, the algorithm remains robust in the presence of stochastic perturbations and observation errors, and it does not require any synchronization between players.
Full work available at URL: https://arxiv.org/abs/1303.2270
Recommendations
- On the robustness of learning in games with stochastically perturbed payoff observations
- A payoff-based learning procedure and its application to traffic games
- Learning in nonatomic games. I: Finite action spaces and population games
- Learning with minimal information in continuous games
- Learning dynamics in games with stochastic perturbations
Stochastic approximation (62L20) Evolutionary games (91A22) Rationality and learning in game theory (91A26)
Cites Work
- A survey on networking games in telecommunications
- Evolutionary Games and Population Dynamics
- Evolutionary dynamics with aggregate shocks
- Potential games
- Optimal properties of stimulus-response learning models.
- Quantal response equilibria for normal form games
- Higher order game dynamics
- Time Average Replicator and Best-Reply Dynamics
- Inertial game dynamics and applications to constrained optimization
- A Simple Adaptive Procedure Leading to Correlated Equilibrium
- Evolutionary Selection in Normal-Form Games
- Exponential weight algorithm in continuous time
- The emergence of rational behavior in the presence of stochastic perturbations
- Two Competing Models of How People Learn in Games
- On the Global Convergence of Stochastic Fictitious Play
- Hessian Riemannian Gradient Flows in Convex Programming
- A payoff-based learning procedure and its application to traffic games
- When can the two-armed bandit algorithm be trusted?
- Attainability of boundary points under reinforcement learning
- Learning through reinforcement and replicator dynamics
- Title not available (Why is that?)
- Learning by trial and error
- Truncated stochastic approximation with moving bounds: convergence
- Individual Q-Learning in Normal Form Games
- Decentralized learning of Nash equilibria in multi-person stochastic games with incomplete information
- Asynchronous stochastic approximation with differential inclusions
- An adjusted payoff-based procedure for normal form games
Cited In (21)
- Continuous-time convergence rates in potential and monotone games
- Replicator dynamics: old and new
- Population games and discrete optimal transport
- Performance of non-cooperative routing over parallel non-observable queues
- Riemannian game dynamics
- Learning in games with continuous action sets and unknown payoff functions
- Robust power modulation for channel state information exchange
- Nested replicator dynamics, nested logit choice, and similarity-based learning
- Inertial game dynamics and applications to constrained optimization
- On best-response dynamics in potential games
- Bifurcation mechanism design -- from optimal flat taxes to better cancer treatments
- Learning in games via reinforcement and regularization
- On the robustness of learning in games with stochastically perturbed payoff observations
- Exploration-exploitation in multi-agent learning: catastrophe theory meets game theory
- On the stability of an adaptive learning dynamics in traffic games
- A stochastic approximation algorithm for stochastic semidefinite programming
- Regularized Bayesian best response learning in finite games
- A unified stochastic approximation framework for learning in games
- No-regret algorithms in on-line learning, games and convex optimization
- Memory loss can prevent chaos in games dynamics
- Analysis of complex dynamics in different bargaining systems
This page was built for publication: Penalty-regulated dynamics and robust learning procedures in games
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3449451)
