On sparsity and overcompleteness in image models.
Pietro Berkes, Richard Turner and Maneesh Sahani
Advances in Neural Information Processing Systems
, Cambridge, USA
Computational models of visual cortex, and in particular those based on sparse
coding, have enjoyed much recent attention. Despite this currency, the question
of how sparse or how over-complete a sparse representation should be, has gone
without principled answer. Here, we use Bayesian model-selection methods to address these questions for a sparse-coding model based on a Student-t prior. Having validated our methods on toy data, we find that natural images are indeed best
modelled by extremely sparse distributions; although for the Student-t prior, the
associated optimal basis size is only modestly over-complete.
|EPrint Type:||Book Section|
|Project Keyword:||Project Keyword UNSPECIFIED|
|Deposited By:||Maneesh Sahani|
|Deposited On:||24 March 2009|