June 4th, 2012 @ 11:30pm Ali Eslami (University of Edinburgh)

The Shape Boltzmann Machine: a Strong Model of Object Shape

Models of the shape of an object play a crucial role in many imaging algorithms, such as those for object detection and segmentation, inpainting and graphics.

Our work addresses the question of how to build a 'strong' probabilistic model of object shape. We define a strong model as one which meets two requirements: 1. Realism – samples from the model look realistic, and 2. Generalization – the model can generate samples that differ from training examples.

We consider a class of models known as Deep Boltzmann Machines (DBMs) and show how a strong model of shape can be constructed using a specific form of DBM which we call the 'Shape Boltzmann Machine' (ShapeBM). We demonstrate that the ShapeBM learns distributions that are qualitatively and quantitatively better than existing models for such tasks.

Joint work with Nicolas Heess and John Winn.

seminars/seminaritems/2012-06-04.txt · Last modified: 2012/05/21 17:36 by silberman