ディジタル磁気記録におけるニューラルネットワーク信号処理方式に関する研究

概要

論文の詳細を見る
Artificial neural networks attempt to simulate the networks of nerve cell of the biological central nervous system. This simulation means a gross cell-by-cell training by use of multi-layer perceptrons. In this report, the multi-layer artificial neural networks using the back propagation learning procedure are introduced as a well-known partial response equalizer. It compares the bit error rate performance of the partial response equalization scheme using the maximum likelihood (ML) detection with that of the maximum a posteriori (MAP) detection. As a result, it is cleared that the optimal number of neurons in the hidden layer are equal for the given three layer networks using ML and MAP detections, respectively.
2009-10-30