文献类型：期刊文献

英文题名：Bit Reduction for Locality-Sensitive Hashing

作者：Liu, Huawen[1];Zhou, Wenhua[2];Zhang, Hong[3];Li, Gang[4];Zhang, Shichao[5];Li, Xuelong[6]

机构：[1]Shaoxing Univ, Dept Comp Sci, Shaoxing 312000, Peoples R China;[2]Jinhua Polytech, Coll Informat Engn, Jinhua 321016, Peoples R China;[3]Shaoxing Univ, Dept Math, Shaoxing 312000, Peoples R China;[4]Deakin Univ, Sch Informat Technol, Burwood, Vic 3125, Australia;[5]Cent South Univ, Sch Comp Sci, Changsha 410083, Peoples R China;[6]Northwestern Polytech Univ, Sch Artificial Intelligence OPt & ElectroNics iOPE, Xian 710072, Peoples R China

年份：0

外文期刊名：IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS

收录：SCI-EXPANDED(收录号：WOS:000972149500001)、、EI(收录号：20231714015494)、Scopus(收录号：2-s2.0-85153393602)、WOS

基金：This work was supported in part by the National Natural Science Foundation of China under Grant 61976195 and in part by NSF of Zhejiang Province under Grant LZ23F020003

语种：英文

外文关键词：Binary codes; Mutual information; Hash functions; Convolutional neural networks; Kernel; Data models; Correlation; Binary representation; hash bit reduction; hash learning; information retrieval; locality-sensitivity hashing (LSH); mutual information; similarity preservation

外文摘要：Locality-sensitive hashing (LSH) has gained ever-increasing popularity in similarity search for large-scale data. It has competitive search performance when the number of generated hash bits is large, reversely bringing adverse dilemmas for its wide applications. The first purpose of this work is to introduce a novel hash bit reduction schema for hashing techniques to derive shorter binary codes, which has not yet received sufficient concerns. To briefly show how the reduction schema works, the second purpose is to present an effective bit reduction method for LSH under the reduction schema. Specifically, after the hash bits are generated by LSH, they will be put into bit pool as candidates. Then mutual information and data labels are exploited to measure the correlation and structural properties between the hash bits, respectively. Eventually, highly correlated and redundant hash bits can be distinguished and then removed accordingly, without deteriorating the performance greatly. The advantages of our reduction method include that it can not only reduce the number of hash bits effectively but also boost retrieval performance of LSH, making it more appealing and practical in real-world applications. Comprehensive experiments were conducted on three public real-world datasets. The experimental results with representative bit selection methods and the state-of-the-art hashing algorithms demonstrate that the proposed method has encouraging and competitive performance.