Learning the Dependency Structure of Latent Factors

In this paper, the authors study latent factor models with dependency structure in the latent space. They propose a general learning framework which induces sparsity on the undirected graphical model imposed on the vector of latent factors. A novel latent factor model SLFA is then proposed as a matrix factorization problem with a special regularization term that encourages collaborative reconstruction. The main benefit (novelty) of the model is that user can simultaneously learn the lower-dimensional representation for data and model the pair-wise relationships between latent factors explicitly. An on-line learning algorithm is devised to make the model feasible for large-scale learning problems. Experimental results on two synthetic data and two real-world data sets demonstrate

Provided by: Georgia Institute of Technology Topic: Software Date Added: Dec 2012 Format: PDF

Find By Topic