File(s) stored somewhere else

Please note: Linked content is NOT stored on Carnegie Mellon University and we can't guarantee its availability, quality, security or accept any liability.

Graph-Based Semi-Supervised Learning as a Generative Model

journal contribution
posted on 01.01.2007, 00:00 by Jingrui He, Jaime G. Carbonell, Yan Liu

This paper proposes and develops a new graph-based semi-supervised learning method. Different from previous graph-based methods that are based on discriminative models, our method is essentially a generative model in that the class conditional probabilities are estimated by graph propagation and the class priors are estimated by linear regression.  Experimental results on various datasets show that the proposed method is superior to existing graph-based semi-supervised learning methods, especially when the labeled subset alone proves insufficient to estimate meaningful class priors.




Usage metrics