Artificial neural networks combined multi-wavelength transmission spectrum feature extraction for sensitive identification of waterborne bacteria

This data set includes the data of (1) the multi-wavelength transmission spectrum of S. aureus suspension with different concentrations, the normalized spectrum of 6 samples and the standard deviation - wavelength curve obtained by variance analysis of the normalized spectral matrix; (2) Typical spectra of five kinds of bacteria at their log phase and their anova curves; (3) The recognition accuracy of validation set obtained by using data of different dimensions as the model input; (4) Recognition accuracies obtained by ANNs(BPNN, GRNN, PNN) models for bacterial samples based on the original spectrum and the characteristic interval spectrum; (5) Recognition accuracies obtained by ANNs(BPNN, GRNN, PNN) models for bacterial samples based on the feature attributes after dimension reduction by PCA.