Evolutionary search framework based on the information geometry

2014 Fiscal Year Final Research Report

• Project/Area Number: 25880012
• Research Category: Grant-in-Aid for Research Activity Start-up
• Allocation Type: Single-year Grants
• Research Field: Soft computing
• Research Institution: Shinshu University
• Principal Investigator: AKIMOTO Yohei 信州大学, 学術研究院工学系, 助教 (20709654)
• Project Period (FY): 2013-08-30 – 2015-03-31
• Keywords: 確率的探索法 / 進化計算 / 情報幾何 / 確率近似 / 高次元最適化 / 制約付き最適化 / ノイズ付き最適化 / 勾配推定

Outline of Final Research Achievements

In this work, we have studied the Information-Geometric Optimization (IGO) that is a recently proposed framework of probability model based search algorithm for arbitrary optimization problems based on the information geometry. We have designed a novel variant of the covariance matrix adaptation evolution strategy for high dimensional continuous optimization, a new algorithm for constraint continuous optimization, both of which are based on the IGO principle. We have also studied the runtime of a general comparison based algorithm with fitness averaging strategy on noisy discrete optimization problems and derived the required number of fitness resampling.

Free Research Field

ソフトコンピューティング