材料設計の新しい手法(GAとニューロの融合)(遺伝的アルゴリズムとニューラルネットワーク)

概要

論文の詳細を見る
The blending design problem is hard to solve in usual methods. Because, this problem is inverse problem of the mapping from composition to physical properties, and which is classified as ill-posed problem. We discuss a new inverse modeling method of neural-network using genetic algorithms, named QIG (Quasi-Inverse model by Genetic algorithms). The QIG minimizes the difference between a target value and the neural-network's output value by searching input space of neural-network. We also argue the possibility to solve blending design problem applying the QIG.
社団法人日本オペレーションズ・リサーチ学会の論文
1994-10-11