Sample Efficient Multiagent Learning in the Presence of Markovian Agents
BücherAngebote / Angebote:
The problem of Multiagent Learning (or MAL) is concerned with thestudy of how intelligent entities can learn and adapt in the presence ofother such entities that are simultaneously adapting. The problem isoften studied in the stylized settings provided by repeated matrix games(a.k.a. normal form games). The goal of this book is to develop MALalgorithms for such a setting that achieve a new set of objectives whichhave not been previously achieved. In particular this book deals withlearning in the presence of a new class of agent behavior that has notbeen studied or modeled before in a MAL context: Markovian agentbehavior. Several new challenges arise when interacting with thisparticular class of agents. The book takes a series of steps towardsbuilding completely autonomous learning algorithms that maximize utilitywhile interacting with such agents. Each algorithm is meticulouslyspecified with a thorough formal treatment that elucidates its keytheoretical properties.
Folgt in ca. 5 Arbeitstagen