The article presents a hypothesis in neuroscience that ganglion cells in the retina are activated by selectively detecting visual features in a scene. This study learns an interpretable graph-based classifier from data to predict the firings of ganglion cells in response to visual stimuli. The classifier defines Mahalanobis distances between graph nodes (visual events) with pre-computed feature vectors, leading to edge weights and a combinatorial graph that is amenable to binary classification. The study aims to extract knowledge from cell firings and provide interpretability to complex neural networks.
Publication date: 4 Jan 2024
Project Page: not provided
Paper: https://arxiv.org/pdf/2401.01813
