Manifold relevance determination
2012 (English)In: Proceedings of the 29th International Conference on Machine Learning, ICML 2012, 2012, 145-152 p.Conference paper (Refereed)
In this paper we present a fully Bayesian latent variable model which exploits conditional non-linear (in)-dependence structures to learn an efficient latent representation. The latent space is factorized to represent shared and private information from multiple views of the data. In contrast to previous approaches, we introduce a relaxation to the discrete segmentation and allow for a "softly" shared latent space. Further, Bayesian techniques allow us to automatically estimate the dimensionality of the latent spaces. The model is capable of capturing structure underlying extremely high dimensional spaces. This is illustrated by modelling unprocessed images with tenths of thousands of pixels. This also allows us to directly generate novel images from the trained model by sampling from the discovered latent spaces. We also demonstrate the model by prediction of human pose in an ambiguous setting. Our Bayesian framework allows us to perform disambiguation in a principled manner by including latent space priors which incorporate the dynamic nature of the data.
Place, publisher, year, edition, pages
2012. 145-152 p.
Bayesian frameworks, Bayesian techniques, Dynamic nature, High dimensional spaces, Human pose, Latent variable models, Multiple views, Private information, Artificial intelligence, Software engineering, Learning systems
IdentifiersURN: urn:nbn:se:kth:diva-105438ScopusID: 2-s2.0-84867113123ISBN: 978-145031285-1OAI: oai:DiVA.org:kth-105438DiVA: diva2:571285
29th International Conference on Machine Learning, ICML 2012, 26 June 2012 through 1 July 2012, Edinburgh
QC 201211222012-11-222012-11-212012-11-22Bibliographically approved