Key points are not available for this paper at this time.
Inspired by the biological olfactory system, this study presents a novel biomimetic olfactory perception model (BOPM) to enhance the odor recognition capabilities of electronic noses (e-noses). It integrates three key components: a biomimetic olfactory epithelium (BOE), biomimetic olfactory bulb (BOB), and biomimetic olfactory cortex (BOC), each mimicking the corresponding biological structures and functions. The BOE employs a neural encoding scheme with multiple wavelet receptive fields to transform raw sensor data into spike information. The BOB reproduces neural circuits for fine-tuning odor signal processing. Finally, the BOC realizes automatic feature learning and classification through a spiking neural network (SNN). The performance of the proposed model was evaluated on two datasets: one collected in our laboratory involving six low-concentration gases and another from the UCI repository containing responses to four gases and their mixtures. The model achieved a recognition accuracy of 95% on the laboratory dataset and 98% on the UCI dataset, significantly outperforming deep learning models such as LeNet-5 and AlexNet. These results validate the model’s ability to leverage key biological mechanisms for enhanced odor detection in e-noses.
Li et al. (Fri,) studied this question.