Figure 6.

Hardware implementation for wine classification. a) Experimental setup used for wine classification. Wine gas was injected into a probe box where the artificial olfactory neuron modules were located. Electrical signals from the artificial olfactory neuron modules were transported to synapses mounted on the printed circuit board (PCB). b) Neural network based on a single‐layer spiking neural network (SNN) for the classification of two wines. The input layers corresponded to the two artificial olfactory neuron modules and the output layers corresponded to the two wines. c) Synapse current collected at the output layer ① (I _syn1) and at the output layer ② (I _syn2). For “Merlot” wine, the spiking frequency of I _syn1 was larger than I _syn2. Otherwise, the result was reversed; i.e., spiking frequency of I _syn2 > I _syn1 for “Shiraz” wine.