University of Cambridge > Talks.cam > Machine Learning @ CUED > Factored Shapes and Appearances for Parts-based Object Understanding AND Transformation Equivariant Boltzmann Machines

Factored Shapes and Appearances for Parts-based Object Understanding AND Transformation Equivariant Boltzmann Machines

Add to your list(s) Download to your calendar using vCal

If you have a question about this talk, please contact Carl Edward Rasmussen.

Two short talks:

  1. Factored Shapes and Appearances for Parts-based Object Understanding

We present a novel generative framework for learning parts-based representations of object classes. Our model, Factored Shapes and Appearances (FSA), employs a highly factored representation to reason about appearance and shape variability across datasets of images. We propose Markov Chain Monte Carlo sampling schemes for efficient inference and learning, and evaluate the model on a number of datasets. Here we consider datasets that exhibit large amounts of variability, both in the shapes of objects in the scene, and in their appearances. We show that the FSA model extracts meaningful parts from training data, and that its parameters and representation can be used to perform a range of tasks, including object parsing, segmentation and fine-grained categorisation.

Joint work with Ali Eslami

  1. Transformation Equivariant Boltzmann Machines

We develop a novel modeling framework for Boltzmann machines, augmenting each hidden unit with a latent transformation assignment variable which describes the selection of the transformed view of the canonical connection weights associated with the unit. This enables the inferences of the model to transform in response to transformed input data in a stable and predictable way, and avoids learning multiple features differing only with respect to the set of transformations. Extending prior work on translation equivariant (convolutional) models, we develop translation and rotation equivariant restricted Boltzmann machines (RBMs) and deep belief nets (DBNs), and demonstrate their effectiveness in learning frequently occurring statistical structure from artificial and natural images.

Joint work with Jyri Kivinen

This talk is part of the Machine Learning @ CUED series.

Tell a friend about this talk:

This talk is included in these lists:

Note that ex-directory lists are not shown.

 

© 2006-2018 Talks.cam, University of Cambridge. Contact Us | Help and Documentation | Privacy and Publicity