Figure 5.

Semi‐empirical software simulation for gas classification. a) Schematic of multi‐layer spiking neural network (SNN) constructed for gas classification. It was composed of two artificial olfactory neurons in the input layer, 100 hidden neurons in two hidden layers each, and four output neurons in the output layer. Two input layers are the neuron module with the SnO₂ gas sensor and the neuron module with the WO₃ gas sensor. Measured spiking frequencies of the two artificial olfactory neuron modules were reflected for the semi‐empirical simulations. b) Flow chart of the simulation for the gas classification. The blue boxes represent semi‐empirical simulations for neuronal operations. c) Recognition rate of the test set and training set as a function of the number of epochs. d) Confusion matrix to show the accuracy of test results. The gas classification of the four gases was performed successfully.