画像処理エキスパートシステムによるマイクロ・キャリア上動物細胞の視覚認識・計測法 : 第4報,GAによるニューラルネットワーク初期値探索問題

概要

論文の詳細を見る
This paper deals with an expert system for the recognition and counting of a mammalian cell, which is the fibroblast, by image processing and a neural network. These cells are cultured on the surface of a microcarrier which is 150-200 micrometers in diameter. This culture method using microcarrier is one of the high efficient productions. To monitor the culture conditions, expert operators must count the number of cells on the microcarrier. We have proposed the counting algorithms, and we have room for improvement with respect to the recognition rate using the neural network. The purpose of this paper is to study the optimization of initial values for a multilayered feed-forward neural network using a genetic algorithm. The genetic algorithm is available for optimization problems. We apply the genetic algorithm to neural network problems, such as the exclusive-or problem, the function approximation problem and the cell recognition problem. We show the validity of this method using the genetic algorithm.
一般社団法人日本機械学会の論文
1993-05-25