ifgovh/sampling_in_gsm

Network model of V1 simple cells which represent a complete probability distribution by a sequence of multivariate stochastic samples. Perceptual inference in a model of natural images is assumed to result in a posterior probability distribution. The image model used is a Gaussian Scale Mixture model that was shown to be both an effective computer vision model for denoising and compression, but also as a good predictor of the mean activity of simple cells in response to various stimuli.

MATLABGPL-3.0