This is a Matlab (MEXed) implementation of the nonparametric max-margin matrix factorization (iPM3F) [1] and its alternative formulation with data augmentation (Gibbs iPM3F) [2]. The code is published on GitHub.

References

1. Minjie Xu, Jun Zhu and Bo Zhang. Nonparametric Max-Margin Matrix Factorization for Collaborative Prediction. In Advances in Neural Information Processing Systems 25, Lake Tahoe, USA, 2012. (NIPS 2012)


2. Minjie Xu, Jun Zhu and Bo Zhang. Fast Max-Margin Matrix Factorization with Data Augmentation. In Proceedings of the 30th International Conference on Machine Learning, Atlanta, USA, 2013. (ICML 2013)

This is a Matlab implementation of the fancy idea by Polson & Scott [1] that reformulates the traditional binary linear SVM problem into a MAP (Maximum a Posteriori) estimation in a probabilistic generative model, and by use of the technique of data augmentation, makes it possible to do very easy and fast probabilistic inference for the solution.
Specifically, I implemented the basic EM algorithm and the MCMC algorithm, as well as the case under spike-and-slab prior. And we extend the algorithm to the Crammer & Singer multi-class SVM [2]!
The code is already highly optimized for Matlab (not further accelerated by MEX though). And it is published on GitHub.

References

1. Nicholas G. Polson and Steven L. Scott. Data augmentation for support vector machines, Bayesian Analysis, 6(1): 1–24, 2011.


2. Hugh Perkins, Minjie Xu, Jun Zhu and Bo Zhang. Fast Parallel SVMs using Data Augmentation. To appear.