CoopHash: Cooperative Learning of Multipurpose Descriptor and Contrastive Pair Generator via Variational MCMC Teaching for Supervised Image Hashing

10/09/2022
by   Khoa D. Doan, et al.
5

Leveraging supervised information can lead to superior retrieval performance in the image hashing domain but the performance degrades significantly without enough labeled data. One effective solution to boost the performance is to employ generative models, such as Generative Adversarial Networks (GANs), to generate synthetic data in an image hashing model. However, GAN-based methods are difficult to train and suffer from mode collapse issue, which prevents the hashing approaches from jointly training the generative models and the hash functions. This limitation results in sub-optimal retrieval performance. To overcome this limitation, we propose a novel framework, the generative cooperative hashing network (CoopHash), which is based on the energy-based cooperative learning. CoopHash jointly learns a powerful generative representation of the data and a robust hash function. CoopHash has two components: a top-down contrastive pair generator that synthesizes contrastive images and a bottom-up multipurpose descriptor that simultaneously represents the images from multiple perspectives, including probability density, hash code, latent code, and category. The two components are jointly learned via a novel likelihood-based cooperative learning scheme. We conduct experiments on several real-world datasets and show that the proposed method outperforms the competing hashing supervised methods, achieving up to 10 over the current state-of-the-art supervised hashing methods, and exhibits a significantly better performance in out-of-distribution retrieval.

READ FULL TEXT
research
04/23/2018

Deep Semantic Hashing with Generative Adversarial Networks

Hashing has been a widely-adopted technique for nearest neighbor search ...
research
04/07/2017

Supervised Deep Hashing for Hierarchical Labeled Data

Recently, hashing methods have been widely used in large-scale image ret...
research
12/12/2016

Deep Supervised Hashing with Triplet Labels

Hashing is one of the most popular and powerful approximate nearest neig...
research
05/17/2021

Prototype-supervised Adversarial Network for Targeted Attack of Deep Hashing

Due to its powerful capability of representation learning and high-effic...
research
01/31/2022

Learning to Hash Naturally Sorts

Locality sensitive hashing pictures a list-wise sorting problem. Its tes...
research
12/17/2022

Hyperbolic Hierarchical Contrastive Hashing

Hierarchical semantic structures, naturally existing in real-world datas...
research
08/20/2021

Unsupervised Domain-adaptive Hash for Networks

Abundant real-world data can be naturally represented by large-scale net...

Please sign up or login with your details

Forgot password? Click here to reset