Abstract:Compact hash code learning has been widely applied to fast similarity search owing to its significantly reduced storage and highly efficient query speed. However, it is still a challenging task to learn discriminative binary codes for perfectly preserving the full pairwise similarities embedded in the high-dimensional real-valued features, such that the promising performance can be guaranteed. To overcome this difficulty, in this paper, we propose a novel scalable supervised asymmetric hashing (SSAH) method, which can skillfully approximate the full-pairwise similarity matrix based on maximum asymmetric inner product of two different non-binary embeddings. In particular, to comprehensively explore the semantic information of data, the supervised label information and the refined latent feature embedding are simultaneously considered to construct the high-quality hashing function and boost the discriminant of the learned binary codes. Specifically, SSAH learns two distinctive hashing functions in conjunction of minimizing the regression loss on the semantic label alignment and the encoding loss on the refined latent features. More importantly, instead of using only part of similarity correlations of data, the full-pairwise similarity matrix is directly utilized to avoid information loss and performance degeneration, and its cumbersome computation complexity on n ×n matrix can be dexterously manipulated during the optimization phase. Furthermore, an efficient alternating optimization scheme with guaranteed convergence is designed to address the resulting discrete optimization problem. The encouraging experimental results on diverse benchmark datasets demonstrate the superiority of the proposed SSAH method in comparison with many recently proposed hashing algorithms.

A Scalable Optimization Mechanism for Pairwise Based Discrete Hashing

Scalable Supervised Asymmetric Hashing with Semantic and Latent Factor Embedding

Nonlinear Discrete Hashing.

Kernel-Based Supervised Discrete Hashing For Image Retrieval

Discrete Double-Bit Hashing

A Fast Optimization Method for General Binary Code Learning

Unsupervised Deep Hashing with Similarity-Adaptive and Discrete Optimization

Sequential Discrete Hashing for Scalable Cross-Modality Similarity Retrieval

Hadamard Codebook Based Deep Hashing.

BATCH: A Scalable Asymmetric Discrete Cross-Modal Hashing

Deep Discrete Hashing with Pairwise Correlation Learning.

Coordinate Discrete Optimization for Efficient Cross-View Image Retrieval.

Hashing With Pairwise Correlation Learning and Reconstruction.

Ordinal Constrained Binary Code Learning for Nearest Neighbor Search.

Asymmetric Similarity-Preserving Discrete Hashing for Image Retrieval

Binary Multidimensional Scaling for Hashing

Learning compact binary code based on multiple heterogeneous features

Supervised Discriminative Hashing For Compact Binary Codes

Enhanced Discrete Multi-modal Hashing: More Constraints Yet Less Time to Learn (extended Abstract)

Towards Optimal Binary Code Learning Via Ordinal Embedding

Learning Efficient Binary Codes from High-Level Feature Representations for Multilabel Image Retrieval