Research on Discrete Hash Algorithm Based on Deep Semantics

2018 2nd annual International Conference on Cloud Technology and Communication Engineering

计算机科学;无线电电子学

Jia, Qinjin^1

College of Engineering, Boston University, 8 St Marys St., Boston

MA

02215, United States^1

关键词: Classified information;    Hash algorithm;    Minimization methods;    Rapid growth;    Semantic information;    Target functions;    Video data;    Video search;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/466/1/012102/pdf
DOI  :  10.1088/1757-899X/466/1/012102

学科分类：计算机科学（综合）

来源: IOP

【 摘 要 】

With the rapid growth of image and video data on the network, hash technology has been widely studied in the field of image and video search in recent years. Benefiting from the latest progress in deep learning, deep hash method has achieved good results in image retrieval. However, the previous deep hash method has the limitation that the semantic information is not fully utilized. In this paper, we develop a discrete hash algorithm based on deep supervision, assuming that learning binary code should be an ideal choice of classification. The pair tag information and classified information are used to learn hash codes within a framework. The output of the last layer is restricted to binary code directly, which is rarely studied in deep hash algorithm. Due to the discrete properties of hash codes, the alternate minimization method is used to optimize the target function. The proposed algorithm is proved to be better than the other supervised hash methods in two public image retrieval databases CIFAR-10 and NUS-WIDE.