2018/06/12 by Fabian Tschopp, Fabian David Tschopp, Michael B. Reiser +4 · 1 voice · 11 citations
Biochemistry, Genetics and Molecular Biology · Computer Science · Engineering · Mathematics · Neuroscience · #Advanced Memory and Neural Computing #Artificial intelligence #Biology #Computer science #Computer vision #Connectome #Convolutional neural network #Functional connectivity #Geometry #Hexagonal crystal system #Insect and Arachnid Ecology and Behavior #Lattice (music) #Mathematics #Neural dynamics and brain function #Neurobiology and Insect Physiology Research #Neuroscience #Orientation (vector space) #Physics #cs.CV #q-bio.NC
paper · pdf · doi:10.48550/arxiv.1806.04793
published in arXiv (Cornell University) (Cornell University) · Work in progress. Final paper with results from an updated model with new connectome data will be coming soon
openalex publication_date 2018/06/12 · arxiv created 2018/06/24 · arxiv updated 2018/06/26 · openalex created_date 2022/10/05 · openalex updated_date 2026/08/04
What can we learn from a connectome? We constructed a simplified model of the first two stages of the fly visual system, the lamina and medulla. The resulting hexagonal lattice convolutional network was trained using backpropagation through time to perform object tracking in natural scene videos. Networks initialized with weights from connectome reconstructions automatically discovered well-known orientation and direction selectivity properties in T4 neurons and their inputs, while networks initialized at random did not. Our work is the first demonstration, that knowledge of the connectome can enable in silico predictions of the functional properties of individual neurons in a circuit, leading to an understanding of circuit function from structure alone.