摘要：Severalhashing-based methods for Approximate Nearest Neighbors (ANN) search in a largedata set have been proposed recently. In particular, semi-supervised hashingutilizes semantic similarity given for a small fraction of pairwise data samplesand active hashing aims to improve the performance for ANN search by relying onan expert for the labeling of the mostinformative points. In this study, we present an active hashing method byprototype-based sample selection. Knowing semantic similarities between clusterprototypes can help extracting relations among the points in the correspondingclusters. For expert labeling, we select prototypes from clusters which do notcontain any data points with labeled information so that all areas canbe covered effectively. Experimental results demonstrate that the proposedactive hashing method improves the performance for ANN search.
