A Projective Simulation Scheme for Partially-Observable Multi-Agent Systems

TL;DR

This paper extends the projective simulation (PS) learning method to partially observable multi-agent systems by incorporating a belief projection operator and observability parameter, supported by theoretical and scenario-based analyses.

Contribution

It introduces a novel partial observability extension to the PS model, including theoretical formulations and network representations for multi-agent scenarios.

Findings

Enhanced PS model handles partial observability

Theoretical framework supports multi-agent scenarios

Network representations demonstrate model effectiveness

Abstract

We introduce a kind of partial observability to the projective simulation (PS) learning method. It is done by adding a belief projection operator and an observability parameter to the original framework of the efficiency of the PS model. I provide theoretical formulations, network representations, and situated scenarios derived from the invasion toy problem as a starting point for some multi-agent PS models.