Hashing for Large-scale Nearest neighbor search: Fatih Cakir (PhD Oral Defense)

10:00 am on Friday, June 20, 2014
12:00 pm on Friday, June 20, 2014
MCS 137
Abstract: Due to the immense growth of multimedia data, expediting similarity search in huge repositories is becoming an increasingly important problem for many applications. One popular solution to expedite the similarity search is to construct hash functions to map the data into a Hamming space where linear search is known to be fast and often sublinear solutions also perform well. These hashing methods can be categorized as unsupervised, semi-supervised and supervised approaches. In this talk, we will discuss the details of three recent representative works from each category, emphasizing the strengths and weaknesses of them. We will additionally discuss open problems in the domain that need to be addressed.