Paper: Exploiting High Dimensional Video Features Using Layered Gaussian Mixture Models Chen, D., Yang, Jie, To appear. The 18th International Conference on Pattern Recognition (ICPR'06), Hong Kong, August 20-24, 2006 Abstract:   Latent Layout Analysis (LLA) is a novel unsupervised learning technique to discover objects in unseen images using a set of un-annotated training images. LLA defines a generative model with two layers of latent variables. The first layer of latent variables define the spatial layout of foreground and background. The latent variables in the second layer define the local appearance of an image. We demonstrate that the proposed LLA signi£cantly outperforms Probabilistic Latent Semantic Analysis (PLSA) in two tasks: object discovery (detection) and object localization. Close