Multi-Dimension Combining (MDC) in abstract Level and Hierarchical MDC (HMDC) to Improve the Classification Accuracy of Enoses

摘要

This paper proposes a new classification algorithm for improving the accuracy of Electronic Noses. The algorithm extends the conventional Multi-Dimension Combining (MDC) of measurement level (PARC method as Multi-layer perceptron, or MLP) into abstract level (PARC methods as K-Nearest Neighbor (KNN), Linear Discriminant Analysis (LDA) and Probabilistic Neural Network (PNN)) and Hierarchical level (HMDC, or Hierarchical Multi-Dimension Combining). The performance of the proposed algorithm is evaluated using experimental data and Enose device of Cyranose 320.