ABSTRACT
We compare a biomimetic model of the early mammalian olfactory system and the olfactory cortex with machine learning methods on odor classification and segmentation problems for input data having biologically plausible characteristics and for polymer sensor recordings. A cortical adaptation-based mechanism for odor segmentation is proposed and context-dependent segmentation is demonstrated in the model. The capability of biomimetic models to integrate several mechanisms in one system may prove highly relevant in addressing complex real-world chemosensory tasks.
