Smart hashing update for fast response
Qiang Yang, Long-Kai Huang, Wei‐Shi Zheng, Yingbiao Ling · 2013
Recent years have witnessed the growing popular-ity of hash function learning for large-scale da-ta search. Although most existing hashing-based methods have achieved promising performance, they are regarded as passive hashing and assume that the labelled pairs are provided in advance. In this paper, we consider updating a hashing mod-el upon gradually increased labelled data in a fast response to users, called smart hashing update (SHU). In order to get a fast response to users, SHU aims to select a small set of hash functions to re-learn and only updates the corresponding hash bits of all data points. More specifically, we put forward two selection methods for performing efficient and effective update. In order to reduce the response time for acquiring a stable hashing code, we also propose an accelerated method to further reduce in-teractions between users and the computer. We e-valuate our proposals on two benchmark data sets. Our experimental results show it is not necessary to update all hash bits in order to adapt the model to new input data, and our model obtains better or similar performance without sacrificing much ac-curacy against the batch mode update. 1