Modeling Images using Transformed Indian Buffet Processes

TitleModeling Images using Transformed Indian Buffet Processes
Publication TypeConference Papers
Year of Publication2012
AuthorsZhai K, Hu Y, Williamson S, Boyd-Graber J
Conference NameInternational Conference of Machine Learning
Date Published2012///
Abstract

Latent feature models are attractive for image modeling, since images generally contain mul- tiple objects. However, many latent feature models ignore that objects can appear at dif- ferent locations or require pre-segmentation of images. While the transformed Indian buffet process (tIBP) provides a method for modeling transformation-invariant features in unsegmented binary images, its current form is inappropriate for real images because of its computational cost and modeling assumptions. We combine the tIBP with likelihoods appropriate for real images and develop an efficient inference, using the cross- correlation between images and features, that is theoretically and empirically faster than existing inference techniques. Our method discovers rea- sonable components and achieve effective image reconstruction in natural images.