The notion of $\alpha$-robustness, a measure of the guaranteed utility each agent retains while pursuing its optimal single agent policy, is introduced, assuming all agents obey the social law.
Abstract
In multi-agent environments, coordinating agents to prevent interference and ensure robust individual performance is a critical challenge. Previous research on social laws for multi-agent systems has primarily focused on deterministic, goal-based settings. This paper extends the concept of social laws to stochastic, reward-based environments, proposing a formalism for defining and verifying their robustness under various conditions. We introduce the notion of $\alpha$-robustness, a measure of the guaranteed utility each agent retains while pursuing its optimal single agent policy, assuming all agents obey the social law. We then present an approach for robustness verification of social laws in stochastic settings, based on a reduction to solving a series of Markov decision processes. Empirical evaluations on toy environments illustrate the potential of our framework.
We study decentralised ergodic control for multi-agent systems operating over stochastic, lossy communication graphs. Each agent follows controlled stochastic dynamics with uniformly elliptic noise on a compact domain and aims to align its occupation measure with a prescribed target distribution π. We separate two regi...
Ramen Ghosh· International Journal of Con...· 1 citation
It is shown that a soft $Q$-learning algorithm converges almost surely to the fixed point of a soft Bellman equation defined by the stationary averages that the behavioral policy and the chosen kernel induce, and what separates the resulting policy from an optimal one.
R. K. Velicheti, Subhonmesh Bose, Tamer Başar· 0 citations
This work introduces a notion of retrospective (backward) counterfactual responsibility, which quantifies an agent's accountability for outcomes resulting from a given strategy profile, and demonstrates how to compute stable strategy profiles in which agents trade off responsibility against expected reward.
Chunyan Mu, Muhammad Najib· Proceedings of the Thirty-Fi...· 0 citations
This work introduces Situational-Constrained Correlated Policy Iteration (SC-CPI), a reinforcement learning algorithm employing a smooth Log-Sum-Exp mechanism for constraint optimization that is the first method learning CE under situational constraints.
Libo Zhang, Zhi-Rui Zeng, Yang Chen et al.· Proceedings of the Thirty-Fi...· 0 citations
This work comprehensively investigates the concept of constant-memory strategies in stochastic games, and uncovers the connection between decision models in single-agent and multi-agent contexts.
Feng-Ming Zhu, Fang-Zhen Lin· Proceedings of the Thirty-Fi...· 0 citations
Standard solution concepts for stochastic games, such as Markov perfect equilibrium and Markov coarse correlated equilibrium, are computationally difficult, and thus, standard decentralized reinforcement-learning algorithms should not generally be expected to converge to them. In this paper, we study the equilibrium ge...
Maurizio D'Andrea, Bar Light· 0 citations
Related blog posts
MIT News · Artificial Intelligence· news.mit.eduOct 13, 2026
Machine-learning networks can learn without detailed instructions from a designer. Those that do may offer insight into how the brain reorganizes itself during learning.
MIT News · Artificial Intelligence· news.mit.eduSep 29, 2026
Professor Sherry Turkle’s new book, “Artificial Intimacy,” offers a withering critique of chatbots and the antisocial dynamics she believes they encourage.
