1相关工作-中国图象图形学报.doc

中图法分类号：TP37 文献标识码：A 文章编号：1006-8961(2014 ) - - 论文引用格式： 主动学习的多标签图像在线分类 在个数据集上用种多标签分类评价指标对本文所提算法进行评估。Multi-Label Image Classification with Active Learning Xu Meixiang1, Sun Fuming1, Li haojie2 1.School of Electronics and Information Engineering, Liaoning University of Technology, Jinzhou, Liaoning, 121001, China; 2.School of Software, Dalian University of Technology, Dalian, Liaoning , 116300, China Abstract: Objective Supervised machine learning methods have been applied to multi-label image classification problems with tremendous success. Despite different learning mechanisms, their performances heavily rely on the number of the labeled training examples. However, the acquisition of labeled examples requires significant efforts from annotators, which hinders the applications of supervised learning methods to large scale problems. In this paper, we extend an active querying method driven by informativeness and representativeness used in single label learning to multi-label image classification. Since the training set grows in active learning, the classifier needs to be updated accordingly. Commonly, the new classifier is required to use all of the training samples for re-training. Under this condition, the training burden of classifier will increase significantly. To speed up the learning efficiency, we explore a highly-effective online learning algorithm. In a nutshell, aiming at the sample selection strategy and the update issues in classifications, a novel algorithm named AIR-OML (Active learning with Informativeness and Representativeness for Online Multi-label Learning) is presented to deal with the massive multi-label classification problems. Method Our sample selection strategy in active learning is based on the min-max theory, querying the most informative and the most representative examples to retrain the multi-label learner. Furthermore, Kullback Leibler Divergence (KLD) and Karush-Kuhn-Tucker (KKT) conditions are utilized to update the multi