作者: Daqi Gao , Zeping Yang , Chaoqian Cai , Fangjun Liu
DOI: 10.1016/J.NEUNET.2012.05.009
关键词:
摘要: This paper studies several types and arrangements of perceptron modules to discriminate quantify multiple odors with an electronic nose. We evaluate the following multilayer perceptron. (A) A single multi-output (SMO) both for discrimination quantification. (B) An SMO followed by (MMO) perceptrons (C) single-output (MSO) (D) MSO quantification, called MSO-MSO model, under conditions: (D1) using a simple one-against-all (OAA) decomposition method; (D2) adopting OAA method virtual balance step; (D3) employing local method, step generalization strategy all together. The experimental results 12 kinds volatile organic compounds at 85 concentration levels in training set 155 test show that model D3 learning procedure is most effective those tested quantification many odors.