Transcript PPT

Similarity Search in Arbitrary Subspaces
Xiang Lian, PhD Candidate, and Lei Chen, Assistant Professor
{xlian, leichen}@cse.ust.hk
Sponsored by MSRA Internet Services Theme Invitation Award
images
feature vectors  In image
query vector

…
…
…
users
subspace
query
query
result
image
index& filter

image database
subspace
databases, each
image is represented by a ddimensional feature vector
A similarity query retrieves
images whose feature vectors
are within distance from a
user-specified query vector
Instead of searching in the
full feature space, our work
aims to answer similarity
queries in arbitrary subspaces
[1] Xiang Lian and Lei Chen. Similarity Search in Arbitrary Subspaces under Lp-Norm. In ICDE, 2008.
[2] Xiang Lian and Lei Chen. A General Cost Model for Dimensionality Reduction in High Dimensional Spaces.
In ICDE, 2007.