YZ Sözlüğü
Yapay Zekanın tam sözlüğü
Team Game Theory
Theoretical framework for cooperative multi-agent learning where agents form a team to achieve a common objective, with shared reward mechanisms and implicit coordination.
Credit Assignment
Fundamental problem in multi-agent learning consisting of correctly assigning reward or blame to each agent for their respective contributions to the team's overall outcome.
Multi-Agent Imitation Learning
Method where agents learn by observing and imitating the behavior of other agents (experts or peers), used to accelerate learning in complex environments with costly exploration.
Multi-Agent Federated Learning
Decentralized approach where agents train local models on their own data and periodically share parameter updates to build a global model without sharing raw data.
Mixed Policies
Strategies in multi-agent learning where each agent can adopt a mix of behaviors (pursuer, evader, cooperater) with changing probabilities depending on the state of the environment and the actions of other agents.
Partial Observation Learning
Paradigm where each agent only has access to a part of the global state of the environment, requiring inference and communication techniques to reconstruct sufficient understanding for decision-making.
Multi-Agent Graph Neural Networks
Deep learning architecture where agents are modeled as nodes in a dynamic graph, allowing to learn representations that capture relationships and dependencies between agents.
Meta-game Learning
Technique where agents learn to learn by quickly adapting to changing strategies of other agents, as in a meta-game where adaptability itself becomes a skill to optimize.
Stabilité Convergente en Apprentissage Multi-Agents
Propriété garantissant que les politiques des agents convergent vers un équilibre stable malgré les interactions continues, condition essentielle pour la fiabilité des systèmes multi-agents déployés.