Multimodal Discovering and Fusion for Semantics Multimedia Analysis

Semantics multimedia. Numerous researches have been focused on multi-modal analysis to detect multimedia semantics. However,there´s still two fundamental issues which have not been adequately addressed. The first is how to choose the best independent modalities.The second is how are a set of modalities optimally fused to map to the high-level semantics. In this paper, statistical and machine learning techniques ISOMAP are applied to solve the two problems. Combining with support vector clustering, ISOMAP are first used to discover independent modalities from raw features. Then, Super Kernel method is applied to optimally fuse the individual modalities. Experiments show that the proposed method can learn multimedia semantics more efficiently than traditional methods.