PROJECT TITLE :

Multi-label dictionary learning for image Annotation. - 2016

ABSTRACT:

Image annotation has attracted a ton of analysis interest, and multi-label learning is an efficient technique for image annotation. The way to effectively exploit the underlying correlation among labels may be a crucial task for multi-label learning. Most existing multi-label learning methods exploit the label correlation only within the output label house, leaving the association between the label and therefore the features of pictures untouched. Although, recently some ways attempt toward exploiting the label correlation in the input feature area by using the label info, they can not effectively conduct the learning method in each the areas simultaneously, and there still exists much room for improvement. During this paper, we propose a unique multi-label learning approach, named multi-label dictionary learning (MLDL) with label consistency regularization and partial-identical label embedding MLDL, which conducts MLDL and partial-identical label embedding simultaneously. In the input feature area, we tend to incorporate the dictionary learning technique into multi-label learning and style the label consistency regularization term to be told the better illustration of options. Within the output label house, we tend to design the partial-identical label embedding, in which the samples with specifically same label set can cluster together, and also the samples with partial-identical label sets can collaboratively represent each different. Experimental results on the 3 widely used image datasets, together with Corel 5K, IAPR TC12, and ESP Game, demonstrate the effectiveness of the proposed approach.