Figure 1
From: Stable Odor Recognition by a neuro-adaptive Electronic Nose
Gas detection using a chemical sensor array and two evaluation methods.
A) 4 × 4 array of chemical sensors. Four sensor classes and 4 sensors per sensor class. B) The pattern vector of the 16 sensor outputs, r1 through r16, are fed in a neural network (upper branch) and then analyzed by the PLS-DA algorithm. For comparison the same data are analyzed without the network (lower branch). C) Example of responses of one sensor class to the five odorants. Upper panel, sequence of stimuli applied. Subsequent panels, overview over the responses of the four sensors of one class to the five stimuli, the average being drawn in bold. Note the prominent jumps upon some stimulus changes. D) Correlation matrix of sensor array response vector in the training phase.
