Content-based image retrieval with the normalized information distance
Computer Vision and Image Understanding
Automatic Image Annotation with Relevance Feedback and Latent Semantic Analysis
Adaptive Multimedial Retrieval: Retrieval, User, and Semantics
Hi-index | 0.00 |
In previous work, we developed a novel Relevance Feedback(RF) framework that learns One-class Support VectorMachines (1SVM) from retrieval experience to represent theset memberships of users' high level semantics. By doing afuzzy classification of a query into the regions of supportrepresented by the 1SVMs, past experience is merged withshort-term (i.e., intra-query) learning. However, this ledto the representation of long-term (i.e., inter-query) learningwith a constantly growing number of 1SVMs in the featurespace. We present an improved version of our earlierwork that uses an incremental k-means algorithm to cluster1SVMs. The main advantage of the improved approach isthat it is scalable and can accelerate query processing byconsidering only a small number of cluster representatives,rather than the entire set of accumulated 1SVMs. Experimentalresults against real data sets demonstrate the effectivenessof the proposed method.