Code for Learning Dependency Graph between Latent Factors from Data
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
src
LICENSE
README.md

README.md

SparseGGM4LatentFactors

Code for Paper, please cite :

@INPROCEEDINGS{yhe12NIPS, title={Learning the Dependency Structure of Latent Factors}, author={Y. He and Y. Qi and K. Kavukcuoglu and H. Park}, booktitle={Proceedings of Advances in Neural Information Processing Systems (NIPS)}, year={2012}, }


Abstract: In this paper, we study latent factor models with dependency structure in the la- tent space. We 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 we can simultaneously learn the lower- dimensional representation for data and model the pairwise 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 that pairwise relationships and latent factors learned by our model provide a more structured way of exploring high-dimensional data, and the learned representations achieve the state-of-the-art classification performance.


Paper PDF @ http://www.cs.cmu.edu/%7Eqyj/papersA08/12-slfa.pdf


Paper supplementary doc @ http://www.cs.cmu.edu/%7Eqyj/papersA08/12-nips-poster.pdf


Related NIPS poster slide @ http://www.cs.virginia.edu/yanjun/paperA14/2012_SLFA_NIPS.pdf