A New Co-training Approach Based on SVM for Image Retrieval

Abstract

It’s difficult to collect vast amounts of labeled data and easy for unlabeled data in collecting image characters. Therefore, it is necessary to define conditions to utilize the unlabeled examples enough. In this paper we present a new co-training approach based on SVM to define two learners, both learners are re-trained after every relevance feedback, and then each of them gives every image in a rank. Experiments show that using co-training idea in CBIR is beneficial, and achieves better performance than some existing methods.