FPGA Implementation of a Probabilistic Neural Network Using Delta-Sigma Modulation for Pattern Discrimination of EMG Signals
This paper proposes a novel probabilistic neural network (PNN) using delta-sigma modulation (DS modulation) with the aim of realizing high performance in the case of the pattern discrimination of bioelectric signals. The proposed net-work includes a statistical model so that the posterior probability for the given input patterns can be estimated. Moreover, the calculation speed of the proposed network in the hardware can be increased since the 1-bit pulse signals with delta-sigma modulators (DSMs) are used for the realization of the internal calculation of the network. In this paper, we implemented the proposed network on a field programmable gate array (FPGA), and discrimination experiments were conducted using the artificial data and the electromyogram (EMG) patterns of an amputee. In the experiments, we confirmed that the proposed network has a high accuracy of pattern discrimination.
Keisuke Shima Toshio Tsuji
Graduate School of Engineering, Hiroshima University Higashi-Hiroshima, Japan 739-8527
国际会议
北京
英文
403-408
2007-05-23（万方平台首次上网日期，不代表论文的发表时间）