指数自然粒子フィルタ
Exponential Natural Particle Filter
自然勾配学習を用いた状態遷移確率で粒子の多様性損失や粒子数増大の問題を緩和する新しい粒子フィルタxNPFを提案し、従来手法より真の目標状態に近く収束することを示した。
著者: Ghazal Zand, Mojtaba Taherkhani, Reza Safabakhsh
分類: cs.LG, cs.NE, cs.RO
原文アブストラクト
Particle Filter algorithm (PF) suffers from some problems such as the loss of particle diversity, the need for large number of particles, and the costly selection of the importance density functions. In this paper, a novel Exponential Natural Particle Filter (xNPF) is introduced to solve the above problems. In this approach, a state transitional probability with the use of natural gradient learning is proposed which balances exploration and exploitation more robustly. The results show that xNPF converges much closer to the true target states than the other state of the art particle filter.
