Self-organising learning of receptive fields in multi-resolution

The statistical likelihood of Gabor filters and primary visual cortex has been of interest for years, yet learning mechanisms proposed did not generate satisfactory Gabor-like receptive fields. In this paper, a new computational model of self-organised Hebbian learning (SOHL) is proposed to work on a multi-resolution image pyramid for the problem of visual receptive field learning. Receptive fields of both orientation and spatial frequency selectivity are observed in the authors' simulation result.