Structured Max Margin Learning on Image Annotation and Multimodal Image Retrieval
Zhen Guo, Zhongfei Mark, P Sandgren Eric, Christos Faloutsos · Multimedia · 2010
In this chapter, we discuss a multimodal framework on image annotation and retrieval based on the max margin approach. The whole problem is mapped to a quadratic programming problem. Our framework is highly scalable in the sense that it takes a constant time to accommodate the database updating without needing to retrain the database from the scratch. The evaluation result shows significant improvements on the performance over a state-of-the-art method.