Slow Learning and Invariance
Speaker: Andreas Maurer , Contact:
Date: November 17 2011
Time: 3:00PM to 4:00PM
Location: Star Sem. Rm D463, Stata
Host: Lorenzo Rosasco, Istituto Italiano di Tecnologia; CBCL, MIT
Kathleen D Sullivan, email@example.comRelevant URL:
Abstract: The meaning of images observed in a natural sequence appears to evolve slowly, with comparatively few sudden changes.
This principle of semantic continuity or slowness has been thought to play an important role in the formation of the mammalian visual cortex, and several machine learning algorithms have been proposed to exploit it for the learning of feature maps invariant under certain geometric transformations, such as translation, rotation or rescaling.
The talk focuses on the relationship between the nature of the observed process and the class of semantic categories or invariances which can be learned from its observation using the slowness principle. Some simulations pertaining to the learning of scale invariance will also be presented.
The Brains & Machines Seminar Series 2011-2012 is being organized by the IIT@MIT lab (a joint lab between MIT and the Italian Institute of Technology.)
See other events that are part of Brains and Machines Seminar Series 2011/2012
See other events happening in November 2011