BiasHash: A Bayesian Hashing Framework for Image Retrieval

MuseHash: Supervised Bayesian Hashing for Multimodal Image Representation

“…In this paper, we therefore focus on hashing methods, particularly aiming to minimise the aforementioned semantic gap of data from different modalities. Hashing approaches are distinguished into single-view [6,27] and multi-view approaches [4,11,20,21,23,[35][36][37]. The former approaches can only handle one modality, while the latter approaches support two or more modalities.…”

“…The former approaches can only handle one modality, while the latter approaches support two or more modalities. In addition, they can categorised into unsupervised [6,11,37] and supervised [4,15,20,21,23,27,35,36] depending on the method of learning hash functions. In general, supervised methods can take more advantage of the inner relationships of data from annotation and due to that they perform better than unsupervised methods.…”

“…Recent works have applied Bayesian frameworks [27,35] to supervised single-view hashing, by adding both a measure of uncertainty and weight regularisation to their predictions. The Bi-asHash method proposed in [27] was shown to outperform stateof-the-art single-view approaches.…”

“…Recent works have applied Bayesian frameworks [27,35] to supervised single-view hashing, by adding both a measure of uncertainty and weight regularisation to their predictions. The Bi-asHash method proposed in [27] was shown to outperform stateof-the-art single-view approaches. In this paper, we extend the BiasHash approach to a multi-view approach, MuseHash, which applies Bayesian regression per modality, uses the sign of the produced Bayesian feature values for binary codes' generation, and fuses the hash-codes into a complete image representation.…”

“…This is done with the use of kernel logistic regression for minimizing their Kullback-Leibler divergence [9]. A recent supervised hashing method, Bayesian Ridge-based Semantic Preserving Hashing (BiasHash) [27], has been proposed for image MuseHash: Supervised Bayesian Hashing for Multimodal Image Representation ACM ICMR '23, June 10-15, 2023, Thessaloniki, Greece retrieval. It inserts a Bayesian framework for learning hash functions and considers that the choice of the training data does not affect the performance of the method.…”