A Parallel Distributed Structure Learning Method of Dynamic Bayesian Networks Using Discrete Particle Swarm Optimization

摘要

本稿では多目的離散Particle Swarm Optimization（DPSO）を用いたDynamic Bayesian Networks（DBN）の並列分散構造学習法を提案する．提案手法は複数のDPSOと多目的遺伝的アルゴリズム（MOGA）による階層構造を持つ．下位層では複数のDPSOがそれぞれDBNの局所的な構造を並列分散的に学習し，上位層ではMOGAがDBN全体の構造学習を行う．数値実験により，提案手法が従来手法と比較してより高速にDBNの構造を学習できることを示す．%In this paper, we present a multi-objective Discrete Particle Optimizer (DPSO) for the parallel distributed learning of Dynamic Bayesian Network (DBN) structures. The proposed method introduces a hierarchical structure consisting of DPSOs and a Multi-Objective Genetic Algorithm (MOGA). Groups of DPSOs in the lower layer find effective DBN sub-network structures and a group of MOGA in the higher layer finds whole of the DBN network structure. Through numerical simulations, we show that the proposed method can find more effective DBN structures and can obtain them faster than the conventional mejhod.