SPLTRAK Abstract Submission
The “Hard Problem” of Olfaction:  Odor Identification from Strongly Occluded Sensory Inputs in a Neuromorphic Olfactory Bulb Circuit
Nabil Imam1 & Thomas A. Cleland2
1Neuromorphic Computing Laboratory, Intel Labs, Santa Clara, CA, United States
2Dept. Psychology, Cornell University, Ithaca, NY, United States

The recognition of meaningful odors encountered within complex and unpredictable chemical environments constitutes the “hard problem” of olfactory neuroscience.  The mammalian olfactory system learns new odors rapidly, exhibits negligible interference among odor memories, and identifies known odors under highly challenging conditions.  The mechanisms by which it does so are unknown.  We here present a general theory for odor learning and identification under noise in the early olfactory system, and demonstrate its efficacy using an olfactory bulb model instantiated in adaptive neuromorphic hardware.  As with biological olfaction, the spike timing-based algorithm utilizes localized computations and resists catastrophic forgetting.  Spike timing-dependent plasticity is employed iteratively over sequential gamma-frequency packets to construct odor representations from the activity of chemosensor arrays mounted in a wind tunnel. Learned odors then are reliably identified even in the presence of strong destructive interference. Noise resistance is enhanced by neuromodulation and contextual priming.  Lifelong learning capabilities are enabled by adult neurogenesis.  The algorithm is applicable to any signal identification problem in which high-dimensional signals are embedded in unknown backgrounds.