Investigator: Jason Castro, Ph.D., Bates College
Project Summary
The goal of this work is to use neural networks to understand how the brain organizes and represents chemical features that confer odor. We will train a graph neural network (GNN) to predict maps of odor-evoked activity in the rat olfactory bulb (OB) from chemical compounds. As the network learns this task, it will in tandem learn molecular attributes that are systematically encoded on the OB’s surface, and which may in turn define the logic of odor perception. This work will provide opportunities for trainees to gain familiarity with methods in data science and machine learning.
