Hashing as Tie-Aware Learning to Rank

05/23/2017
by   Kun He, et al.
0

Hashing, or learning binary embeddings of data, is frequently used in nearest neighbor retrieval. In this paper, we develop learning to rank formulations for hashing, aimed at directly optimizing ranking-based evaluation metrics such as Average Precision (AP) and Normalized Discounted Cumulative Gain (NDCG). We first observe that the integer-valued Hamming distance often leads to tied rankings, and propose to use tie-aware versions of AP and NDCG to evaluate hashing for retrieval. Then, to optimize tie-aware ranking metrics, we derive their continuous relaxations, and perform gradient-based optimization with deep neural networks. Our results establish the new state-of-the-art for image retrieval by Hamming ranking in common benchmarks.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
03/02/2018

Hashing with Mutual Information

Binary vector embeddings enable fast nearest neighbor retrieval in large...
research
09/15/2023

Optimization of Rank Losses for Image Retrieval

In image retrieval, standard evaluation metrics rely on score ranking, a...
research
07/04/2014

Optimizing Ranking Measures for Compact Binary Code Learning

Hashing has proven a valuable tool for large-scale information retrieval...
research
06/09/2011

Ranking via Sinkhorn Propagation

It is of increasing importance to develop learning methods for ranking. ...
research
02/14/2020

Learning to rank for uplift modeling

Uplift modeling has effectively been used in fields such as marketing an...
research
07/05/2022

Hierarchical Average Precision Training for Pertinent Image Retrieval

Image Retrieval is commonly evaluated with Average Precision (AP) or Rec...
research
08/31/2020

Optimize What You Evaluate With: A Simple Yet Effective Framework For Direct Optimization Of IR Metrics

Learning-to-rank has been intensively studied and has shown significantl...

Please sign up or login with your details

Forgot password? Click here to reset