Supervised Dimension Reduction with Topic Models


Khoat Than, Tu Bao Ho, Duy Khuong Nguyen, Ngoc Khanh Pham ;
Proceedings of the Asian Conference on Machine Learning, PMLR 25:395-410, 2012.


We consider supervised dimension reduction (SDR) for problems with discrete variables. Existing methods are computationally expensive, and often do not take the local structure of data into consideration when searching for a low-dimensional space. In this paper, we propose a novel framework for SDR which is (1) general and fiexible so that it can be easily adapted to various unsupervised topic models, (2) able to inherit scalability of unsupervised topic models, and (3) can exploit well label information and local structure of data when searching for a new space. Extensive experiments with adaptations to three models demonstrate that our framework can yield scalable and qualitative methods for SDR. One of those adaptations can perform better than the state-of-the-art method for SDR while enjoying significantly faster speed.

Related Material