Meta Cross-Modal Hashing on Long-Tailed Data

TL;DR

This paper introduces MetaCMH, a meta-learning based cross-modal hashing method designed to effectively handle long-tailed data distributions, improving retrieval performance especially for rare classes in multi-modal datasets.

Contribution

MetaCMH is the first to integrate meta-learning with cross-modal hashing to address long-tailed data challenges, combining direct and memory features for better tail class representation.

Findings

MetaCMH outperforms state-of-the-art methods on long-tailed datasets.

It significantly improves retrieval accuracy for tail classes.

The method effectively balances features for head and tail classes.

Abstract

Due to the advantage of reducing storage while speeding up query time on big heterogeneous data, cross-modal hashing has been extensively studied for approximate nearest neighbor search of multi-modal data. Most hashing methods assume that training data is class-balanced.However, in practice, real world data often have a long-tailed distribution. In this paper, we introduce a meta-learning based cross-modal hashing method (MetaCMH) to handle long-tailed data. Due to the lack of training samples in the tail classes, MetaCMH first learns direct features from data in different modalities, and then introduces an associative memory module to learn the memory features of samples of the tail classes. It then combines the direct and memory features to obtain meta features for each sample. For samples of the head classes of the long tail distribution, the weight of the direct features is larger, because there are enough training data to learn them well; while for rare classes, the weight of the memory features is larger. Finally, MetaCMH uses a likelihood loss function to preserve the similarity in different modalities and learns hash functions in an end-to-end fashion. Experiments on long-tailed datasets show that MetaCMH performs significantly better than state-of-the-art methods, especially on the tail classes.