Image classification using a set of labeled and unlabeled and unlabeled images

摘要

Image classification into meaningufl classes is essentially a supervised pattern recognition problem. These classes include indoor, outdoor, landscape, urban, faces, etc The recognition problem necessitates a large set of labeled examples for training the classifier. Any stratagem, which reduces the burden of labeling, is thereforme very important to the deployment of such classifiers in practical applications. In this papr we show that the labeled training set can be augmented by an unlableled set of examples in order to boost the performance of the classifer. In general, the set of unlabeled examples is not guaranteed to improve the classifier performance. We show that if the actual example to be lableled are automatically selected through an unsupervised clustering step, the performance is more likely to improve with the unlabeled set. In this paper, we first present a modified EM algorithm, which combines labeled and unlabeled sets for training. We then apply this algorithm to image classification. Using mutually excusive classes we show that the clustering step is crucial to the improvement in classifer performance.