Dictionary ensemble based multi instance active learning method for image categorization
Gökhan Koçyiğit, Yusuf Yaslan · 2016
Multiple Instance Learning (MIL) has become a prominent framework for image categorization problem. In MIL framework, images are described as the combination of multiple regions. Active learning in MIL framework becomes useful wregions. Active learning in MIL framework becomes useful when large amount of unlabeled data is available and labeling is too costly to handle for each unlabeled data. In the literature, there are some researches on MI active learning but none of them take advantage of the ensemble techniques and sparse coding. In this work, we study a Dictionary Ensemble based MI Active Learning method. Experiments show that the proposed algorithm has higher classification accuracy over otherhen large amount of unlabeled data is available and labeling is too costly to handle for each unlabeled data. In the literature, there are some researches on MI active learning but none of them take advantage of the ensemble techniques and sparse coding. In this work, we study a Dictionary Ensemble based MI Active Learning method. Experiments show that the proposed algorithm has higher classification accuracy over other techniques.