TY - GEN
T1 - Dimension-specific search for multimedia retrieval
AU - Huang, Zi
AU - Shen, Heng Tao
AU - Song, Dawei
AU - Li, Xue
AU - Rueger, Stefan
PY - 2009
Y1 - 2009
N2 - Observing that current Global Similarity Measures (GSM) which average the effect of few significant differences on all dimensions may cause possible performance limitation, we propose the first Dimension-specific Similarity Measure (DSM) to take local dimensionspecific constraints into consideration. The rationale for DSM is that significant differences on some individual dimensions may lead to different semantics. An efficient search algorithm is proposed to achieve fast Dimension-specific KNN (DKNN) retrieval. Experiment results show that our methods outperform traditional methods by large gaps.
AB - Observing that current Global Similarity Measures (GSM) which average the effect of few significant differences on all dimensions may cause possible performance limitation, we propose the first Dimension-specific Similarity Measure (DSM) to take local dimensionspecific constraints into consideration. The rationale for DSM is that significant differences on some individual dimensions may lead to different semantics. An efficient search algorithm is proposed to achieve fast Dimension-specific KNN (DKNN) retrieval. Experiment results show that our methods outperform traditional methods by large gaps.
UR - https://www.scopus.com/pages/publications/67650178264
U2 - 10.1007/978-3-642-00887-0_60
DO - 10.1007/978-3-642-00887-0_60
M3 - Conference contribution
AN - SCOPUS:67650178264
SN - 9783642008863
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 693
EP - 698
BT - Database Systems for Advanced Applications - 14th International Conference, DASFAA 2009, Proceedings
T2 - 14th International Conference on Database Systems for Advanced Applications, DASFAA 2009
Y2 - 21 April 2009 through 23 April 2009
ER -