The field of multi-agent systems and game theory is rapidly evolving, with a focus on developing innovative solutions to complex problems. Recent research has explored the application of game-theoretic frameworks to federated learning, highlighting the importance of incentive alignment and cooperation among heterogeneous agents. Additionally, there has been significant progress in the development of algorithms for multi-agent reinforcement learning, including the use of optimism as a risk-seeking objective and the introduction of decentralized asynchronous multi-player bandits. Other notable advancements include the achievement of Pareto optimality in games via single-bit feedback and the development of efficient approximation algorithms for fair influence maximization under maximin constraint. Noteworthy papers include 'Incentives in Federated Learning with Heterogeneous Agents', which introduces a game-theoretic framework for capturing heterogeneous data, and 'Learning from Delayed Feedback in Games via Extra Prediction', which proposes a weighted Optimistic Follow-the-Regularized-Leader algorithm for overcoming discrepancies in optimization among agents.
Advances in Multi-Agent Systems and Game Theory
Sources
Incentives in Federated Learning with Heterogeneous Agents
Learning from Delayed Feedback in Games via Extra Prediction
Optimism as Risk-Seeking in Multi-Agent Reinforcement Learning
Asynchronous Policy Gradient Aggregation for Efficient Distributed Reinforcement Learning
Model-Free Dynamic Consensus in Multi-Agent Systems: A Q-Function Perspective
Decentralized Asynchronous Multi-player Bandits
Achieving Pareto Optimality in Games via Single-bit Feedback
Efficient Approximation Algorithms for Fair Influence Maximization under Maximin Constraint
Cooperation in Bilateral Generalized Network Creation
Multi-Agent Stage-wise Conservative Linear Bandits
Online Minimization of Polarization and Disagreement via Low-Rank Matrix Bandits
Exploring one-dimensional, binary, radius-2 cellular automata, over cyclic configurations, in terms of their ability to solve decision problems by distributed consensus
Learning to Play Multi-Follower Bayesian Stackelberg Games
Adversarial Social Influence: Modeling Persuasion in Contested Social Networks
Stability and Robustness of Time-Varying Opinion Dynamics: A Graph-Theoretic Approach
Incentive Analysis of Collusion in Fair Division
A Linear Programming Approach to Estimate the Core in Cooperative Games