Supervised Discrete Hashing Research Articles

Jointly sparse fast hashing with orthogonal learning for large-scale image retrieval

Hash learning is a hot topic since it can save storage space and perform fast retrieval. One of the most representative hashing methods is Supervised Discrete Hashing (SDH). However, there exist several problems in SDH. First, the potential of sparse feature extraction has been overlooked in the SDH-based methods. Second, SDH is incapable of preventing large information loss between the binary codes and the features. Third, the discrete cyclic coordinate descent (DCC) method is time-consuming. To address the aforementioned issues, we propose a novel regression model for hash learning named Jointly Sparse Fast Hashing (JSFH). By designing an orthogonal transformation, we introduce the jointly sparse regression into hash learning. The proposed method successfully integrates the feature extraction and hash learning into a unified framework, the original data and the label information can be comprehensively utilized such that the binary code generated by our method can be more discriminative with less information loss. Furthermore, we adopt an iterative algorithm for optimization, the optimization problem has closed-form solutions for each variable such that the designed iterative algorithm is highly efficient compared to the classic supervised hashing methods. The experimental results on four large-scale datasets demonstrate its superior performance.

Two-Stage Supervised Discrete Hashing for Cross-Modal Retrieval

Recently, hashing-based multimodal learning systems have received increasing attention due to their query efficiency and parsimonious storage costs. However, impeded by the quantization loss caused by numerical optimization, the existing cross-media hashing approaches are unable to capture all the discriminative information present in the original multimodal data. Besides, most cross-modal methods belong to the one-step paradigm, which learn the binary codes and hash function simultaneously, increasing the complexity of optimization. To address these issues, we propose a novel two-stage approach, named the two-stage supervised discrete hashing (TSDH) method. In particular, in the first phase, TSDH generates a latent representation for each modality. These representations are then mapped to a common Hamming space to generate the binary codes. In addition, TSDH directly endows the hash codes with the semantic labels, enhancing the discriminatory power of the learned binary codes. A discrete hash optimization approach is developed to learn the binary codes without relaxation, avoiding the large quantization loss. The proposed hash function learning scheme reuses the semantic information contained by the embeddings, endowing the hash functions with enhanced discriminability. Extensive experiments on several databases demonstrate the effectiveness of the developed TSDH, outperforming several recent competitive cross-media algorithms.

Robust supervised discrete hashing

In this paper, we proposed a more robust supervised hashing framework based on the Cauchy loss function and Supervised Discrete Hashing (SDH) called Robust Supervised Discrete Hashing (RSDH), which can learn a robust subspace consisted of binary codes. The Cauchy loss is used to measure the error between the label matrix and the product of the decomposed matrices. RSDH can not only reduce the outliers and noise of the hashing codes, but also achieve the more satisfactory retrieval effect. Image retrieval experiments demonstrate that RSDH performs better than the other hashing methods.

Deep GAN-Based Cross-Spectral Cross-Resolution Iris Recognition

In recent years, cross-spectral iris recognition has emerged as a promising biometric approach to establish the identity of individuals. However, matching iris images acquired at different spectral bands (i.e., matching a visible (VIS) iris probe to a gallery of near-infrared (NIR) iris images or vice versa) shows a significant performance degradation when compared to intraband NIR matching. Hence, in this paper, we have investigated a range of deep convolutional generative adversarial network (DCGAN) architectures to further improve the accuracy of cross-spectral iris recognition methods. Moreover, unlike the existing works in the literature, we introduce a resolution difference into the classical cross-spectral matching problem domain. We have developed two different novel techniques using the conditional generative adversarial network (cGAN) as a backbone architecture for cross-spectral iris matching. In the first approach, we simultaneously address the cross-resolution and cross-spectral matching problem by training a cGAN that jointly translates cross-resolution as well as cross-spectral tasks to the same resolution and within the same spectrum. In the second approach, we design a coupled generative adversarial network (cpGAN) architecture consisting of a pair of cGAN modules that project the VIS and NIR iris images into a low-dimensional embedding domain to ensure maximum pairwise similarity between the feature vectors from the two iris modalities of the same subject. To assure the efficacy of our methods, we perform several experiments considering multiple real-life scenarios on three publicly-available cross-spectral iris datasets. Our best experimental results obtained from the cpGAN network outperform the existing benchmark convolutional neural network (CNN) with a supervised discrete hashing (SDH) approach <xref ref-type="bibr" rid="ref1" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">[1]</xref> by as much as 1.67%, and 2.22% GAR at FAR of 0.01, while our cGAN provides recognition accuracy with significantly lower EER value of 1.5%, and 1.54% for PolyU bi-spectral dataset, and Cross-eyed-cross-spectral iris recognition database, respectively. It indicates the superiority of our approaches over results previously published in the literature.

Distributed Supervised Discrete Hashing With Relaxation

The data-dependent hash methods are becoming more and more attractive because they perform well in fast retrieval and storing high-dimensional data. Most existing supervised hashes are centralized, such as supervised discrete hashing (SDH) and supervised discrete hashing with relaxation (SDHR). The SDH algorithm determines the regression target by using ordinary least squares regression and the traditional zero-one matrix encoding of class label information. And SDHR is a constraint to the regression target matrix so that each example is correctly classified and satisfies a larger margin, so as to achieve the purpose of optimizing the regression target. In real environment, a large amount of data will be distributed in different nodes. Therefore, the centralized hash method has great limitations. In this article, we propose distributed supervised discrete hashing algorithm with relaxation (DSDHR) based on SDHR. The SDHR algorithm is introduced into the distributed network. In this framework, all nodes share a centralized hash learning model. At the same time, in order to ensure that the distributed hash algorithm is updated in parallel on multiple nodes, consistency constraints are introduced. In each node, alternative iterative methods are used to obtain the binary hash code, regression target and hash function. Experiments show that DSDHR has certain competitive advantages over some centralized hashing algorithms and some distributed hashing algorithms.

Relaxed Locality Preserving Supervised Discrete Hashing

Supervised discrete hashing (SDH) has widely attracted attention because it can directly obtain hash codes rather than relaxing the discrete constraint, which reduces information loss during the quantization process. However, SDH and its extensions tend to ignore the underlying geometry structure of data, while in many cases, data is lying on a low-dimensional manifold. Therefore, based on SDH, we propose a novel manifold learning-based supervised discrete hashing method in this paper to further improve the performance. Our goal is to learn compact hash codes by discovering latent information and using the local structure of data. To this end, we utilize the anchor graph to capture local neighborhood relationship embedding in data for manifold learning. Besides, since the previous methods use a strict binary label matrix for classification, we jointly optimize a relaxed label matrix to increase the discriminative ability of the proposed model. As such, the learned hash codes are more suitable for classification. Compared with SDH and other popular hashing methods, the experimental results on four large-scale datasets show that the proposed hashing method has better performance.

A General Framework for Deep Supervised Discrete Hashing

With the rapid growth of image and video data on the web, hashing has been extensively studied for image or video search in recent years. Benefiting from recent advances in deep learning, deep hashing methods have shown superior performance over the traditional hashing methods. However, there are some limitations of previous deep hashing methods (e.g., the semantic information is not fully exploited). In this paper, we develop a general deep supervised discrete hashing framework based on the assumption that the learned binary codes should be ideal for classification. Both the similarity information and the classification information are used to learn the hash codes within one stream framework. We constrain the outputs of the last layer to be binary codes directly, which is rarely investigated in deep hashing algorithms. Besides, both the pairwise similarity information and the triplet ranking information are exploited in this paper. In addition, two different loss functions are presented: $${l_2}$$ loss and hinge loss, which are carefully designed for the classification term under the one stream framework. Because of the discrete nature of hash codes, an alternating minimization method is used to optimize the objective function. Experimental results have shown that our approach outperforms current state-of-the-art methods on benchmark datasets.

Distributed Fast Supervised Discrete Hashing

Hash-based learning has attracted considerable attention due to its fast retrieval speed and low computational cost for the large-scale database. Compared with unsupervised hashing, supervised hashing achieves higher retrieval accuracy generally by leveraging supervised information. Most existing supervised hashing methods, such as supervised discrete hashing (SDH) and fast SDH (FSDH), are concerned more with the centralized setting. SDH regresses the hash code to its corresponding label, rather FSDH regressing each label to its corresponding hash code. However, in many realistic applications, large amounts of data are usually distributed across different sites. Thus, supervised distributed hashing (SupDisH), which is based on the distributed framework and supervised learning, has been proposed and liberates the limitations of centralized hashing. In this paper, based on FSDH, we propose the distributed fast supervised discrete hashing (DFSDH), which both inherits the excellent retrieval performance of SupDisH and gets significant enhancement in efficiency. Specifically, FSDH is introduced into a distributed framework, in which the centralized hash learning model is shared by all agents. Meanwhile, consistency constraints are introduced to ensure that multiple agents deal with distributed hash learning in parallel. For each agent, an alternate iterative procedure is employed to obtain high-quality binary codes and hashing function. The extensive experiments demonstrate that DFSDH is competitive to most centralized supervised hashing methods and existing distributed hashing methods.

Hadamard Coding for Supervised Discrete Hashing.

In this paper, we propose a learning-based supervised discrete hashing method. Binary hashing is widely used for large-scale image retrieval as well as video and document searches because the compact binary code representation is essential for data storage and reasonable for query searches using bit-operations. The recently proposed supervised discrete hashing (SDH) method efficiently solves mixed-integer programming problems by alternating optimization and the discrete cyclic coordinate descent (DCC) method. Based on some preliminary experiments, we show that the SDH method can be simplified without performance degradation. We analyze the simplified model and provide a mathematically exact solution thereof; we reveal that the exact binary code is provided by a "Hadamard matrix." Therefore, we named our method Hadamard codedsupervised discrete hashing (HC-SDH). In contrast to SDH, our model does not require an alternating optimization algorithm and does not depend on initial values. HC-SDH is also easier to implement than iterative quantization (ITQ). Experimental results involving a large-scale database show that Hadamard coding outperforms conventional SDH in terms of precision, recall, and computational time. On the large datasets SUN-397 and ImageNet, HC-SDH provides a superior mean average of precision (mAP) and top-accuracy compared to the conventional SDH methods with the same code length and FastHash. The training time of HC-SDH is 170 times faster than conventional SDH and the testing time including the encoding time is seven times faster than FastHash which encodes using a binary-tree.

Supervised hashing with adaptive discrete optimization for multimedia retrieval

Hashing techniques show significant advantage in dealing with enormous high-dimensional image and multimedia data. Specifically, learning based hashing methods attract a lot of attention from researchers thanks to its great performance in image retrieval. But discrete constraint problem of learning based hashing methods makes the optimization extremely difficult, which can be shown to be NP hard. Thus, most of learning based hashing methods relax the constraint and get a suboptimal result. Recently, some researchers propose discrete optimization hashing techniques to learn hash bits without any relaxation and achieve promising results. But, discrete optimization hashing method like Supervised Discrete Hashing (SDH) roughly renews all binary codes and leads to a time-consuming problem. In this paper, we propose an adaptive discrete cyclic coordinate descent (ACC) method to effectively solve discrete optimization problem. The specific objective of our study is to boost the efficiency of discrete hash optimization with equivalent performance. We evaluate the proposed method on image and multimedia databases: CIFAR-10, NUS-WIDE and MIRFLickr-25k and show that our method achieves speed-up over compared the state-of-the-art methods, while having on-par and in some cases even better performance.

Supervised Discrete Hashing With Relaxation.

Data-dependent hashing has recently attracted attention due to being able to support efficient retrieval and storage of high-dimensional data, such as documents, images, and videos. In this paper, we propose a novel learning-based hashing method called "supervised discrete hashing with relaxation" (SDHR) based on "supervised discrete hashing" (SDH). SDH uses ordinary least squares regression and traditional zero-one matrix encoding of class label information as the regression target (code words), thus fixing the regression target. In SDHR, the regression target is instead optimized. The optimized regression target matrix satisfies a large margin constraint for correct classification of each example. Compared with SDH, which uses the traditional zero-one matrix, SDHR utilizes the learned regression target matrix and, therefore, more accurately measures the classification error of the regression model and is more flexible. As expected, SDHR generally outperforms SDH. Experimental results on two large-scale image data sets (CIFAR-10 and MNIST) and a large-scale and challenging face data set (FRGC) demonstrate the effectiveness and efficiency of SDHR.