Big Data

Similarity Estimation Using Bayes Ensembles

Date Added: Apr 2010
Format: PDF

Learning similarity functions is an important task for multimedia retrieval and data mining. In data mining, distance measures can be used in various algorithms for classification and clustering. To improve classification, the learned distance measure can be plugged into any instance-based learner like kNN classification. Though clustering is basically an unsupervised problem, learning a similarity function on a small set of manually annotated objects is often enough to guide clustering algorithms to group semantically more similar objects. For similarity search, adaptive similarity measures provide a powerful method to bridge the semantic gap between feature representations and user expectations.