Visual Scene Representation with Hierarchical Equivariant Sparse Coding

Shewmake, C., Buracas, D., Lillemark, H., Shin, J., Bekkers, E., Miolane, N., Olshausen, B.

A hierarchical neural network is proposed for unsupervised learning of part-whole scene decompositions, achieving hierarchical equivariance via Bayesian inference to mimic visual cortex-like perceptual processing.