ResearchTrend.AI
  • Papers
  • Communities
  • Events
  • Blog
  • Pricing
Papers
Communities
Social Events
Terms and Conditions
Pricing
Parameter LabParameter LabTwitterGitHubLinkedInBlueskyYoutube

© 2025 ResearchTrend.AI, All rights reserved.

  1. Home
  2. Papers
  3. 2210.05899
25
2

A Lower Bound of Hash Codes' Performance

12 October 2022
Xiaosu Zhu
Jingkuan Song
Yu Lei
Lianli Gao
Hengtao Shen
ArXivPDFHTML
Abstract

As a crucial approach for compact representation learning, hashing has achieved great success in effectiveness and efficiency. Numerous heuristic Hamming space metric learning objectives are designed to obtain high-quality hash codes. Nevertheless, a theoretical analysis of criteria for learning good hash codes remains largely unexploited. In this paper, we prove that inter-class distinctiveness and intra-class compactness among hash codes determine the lower bound of hash codes' performance. Promoting these two characteristics could lift the bound and improve hash learning. We then propose a surrogate model to fully exploit the above objective by estimating the posterior of hash codes and controlling it, which results in a low-bias optimization. Extensive experiments reveal the effectiveness of the proposed method. By testing on a series of hash-models, we obtain performance improvements among all of them, with an up to 26.5%26.5\%26.5% increase in mean Average Precision and an up to 20.5%20.5\%20.5% increase in accuracy. Our code is publicly available at https://github.com/VL-Group/LBHash.

View on arXiv
Comments on this paper