Generalized mixture models, semi-supervised learning, and unknown class inference

Generalized mixture models, semi-supervised learning, and unknown class inference

In this paper, we discuss generalized mixture models and related semi-supervised learning methods, and show how they can be used to provide explicit methods for unknown class inference. After a brief description of standard mixture modeling and current model-based semi-supervised learning methods, we provide the generalization and discuss its computational implementation using three-stage expec...

Semi-Supervised Learning of Mixture Models

This paper analyzes the performance of semisupervised learning of mixture models. We show that unlabeled data can lead to an increase in classification error even in situations where additional labeled data would decrease classification error. We present a mathematical analysis of this “degradation” phenomenon and show that it is due to the fact that bias may be adversely affected by unlabeled ...

Boosting Mixture Models for Semi-supervised Learning

Semi-Supervised Learning of Mixture Models and Bayesian Networks

This paper analyzes the performance of semisupervised learning of mixture models. We show that unlabeled data can lead to an increase in classification error even in situations where additional labeled data would decrease classification error. This behavior contradicts several empirical results reported in the literature. We present a mathematical analysis of this “degradation” phenomenon and s...

Multi-Instance Mixture Models and Semi-Supervised Learning

Multi-instance (MI) learning is a variant of supervised learning where labeled examples consist of bags (i.e. multi-sets) of feature vectors instead of just a single feature vector. Under standard assumptions, MI learning can be understood as a type of semisupervised learning (SSL). The difference between MI learning and SSL is that positive bag labels provide weak label information for the ins...