Deep hashing for compact binary codes learning

By Venice Erin Liong, Jiwen Lu, Gang Wang, Pierre Moulin, Jie Zhou. Computer Vision and Pattern Recognition, 2015

550 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/5832724acb9606ad7d34f8a92acf51e78b935d75

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

Binary Stochastic Representations for Large Multi-class Classification (2017, background): “As noted by [18], the compactness of the code is crucial: a larger code ensures a better precision measure and less false positive, but at the same time decreases the recall and more false negative are retrieved.”
Deep Sketch Hashing: Fast Free-Hand Sketch-Based Image Retrieval (2017, methodology): “Related Work Hashing techniques [42, 16, 34, 62, 39, 30, 55, 35, 17, 14, 74, 36, 14, 70, 37, 47, 38, 54, 25, 33, 41] have recently been successfully applied to encode highdimensional features into compact similarity-preserving binary codes, which enables extremely fast similarity search by the use o”
HashNet: Deep Learning to Hash by Continuation (2017, background): “Recently, deep learning to hash methods [37, 20, 31, 8, 40] have shown that end-to-end learning of feature representation and hash coding can be more effective using deep neural networks [18, 2], which can naturally encode any nonlinear hash functions.”
Feature Learning Based Deep Supervised Hashing with Pairwise Labels (2015, methodology): “…[Zhao et al., 2015a; Lai et al., 2015; Zhang et al., 2015] Although a lot of hashing methods have been proposed as shown above, most existing hashing methods, including some deep hashing methods [Salakhutdinov and Hinton, 2009; Masci et al., 2014; Liong et al., 2015], are based on ar X iv :1 51 1.”
Learning Discrete Representations via Information Maximizing Self-Augmented Training (2017, methodology): “Erin Liong et al. (2015) recently proposed to use deep neural networks for the mapping function and achieved stateof-the-art results.”
Accelerating Large-Scale Inference with Anisotropic Vector Quantization (2019, methodology): “Quantization methods of this form are available for binary quantization (He et al., 2013; Liong et al., 2015; Dai et al., 2017),”
Binary Generative Adversarial Networks for Image Retrieval (2017, methodology): “On the other hand, the performance of deep hashing methods (DeepBit [28] and DH [7]) is not superior.”
Zero-Shot Sketch-Image Hashing (2018, methodology): “When testing the baseline of bit regularization similar to [37], we experience an unstable training procedure easily leading to overfit10 20 30 40 50 60 70 80 90 100 0.”
Variational Deep Semantic Hashing for Text Documents (2017, background): “Most of the related work has focused on image data [4, 16, 19, 39] rather than text documents probably due to the effectiveness of the convolution neural networks (CNNs) to learn good low-dimensional representations of images.”
Unsupervised Deep Hashing with Similarity-Adaptive and Discrete Optimization (2018, methodology): “These unsupervised models are trained by minimizing either the quantization loss [23], [25] or data reconstruction error [6], which fail to explicitly preserve the data similarities in the learned binary codes and thus achieve far below satisfactory performance.”
Learning to Hash for Indexing Big Data—A Survey (2015, methodology): “, kernel-based supervised hashing (KSH) [22], in terms of search accuracy on some image data sets [99]; the deep hashing method and its supervised version are slightly better than ITQ and its supervised version CCA þ ITQ, respectively [106], [111].”
Deep Supervised Hashing for Fast Image Retrieval (2016, methodology): “To overcome this limitation, Supervised Hashing with Kernels (KSH) [17] and Binary Reconstructive Embedding (BRE) [13] are proposed to learn similaritypreserving hashing functions in kernel space; Deep Hashing (DH) [3] exploits a non-linear deep network to produce binary codes.”
Deep Cross-Modal Hashing (2016, methodology): “There also exist some methods which adopt deep learning for uni-modal hashing (Zhao et al., 2015; Liong et al., 2015).”
Person Re-identification: Past, Present and Future (2016, background): “Another application of supervised hashing consists of image retrieval [152], [153], [154], [155], which is the interest of this section.”
Progressive Generative Hashing for Image Retrieval (2018, background): “DH [Liong et al. , 2015] and Deepbit [Lin et al. , 2016] serve as the pioneering work that learnt the nonlinear end-to-end hash function under independent and balanced constraints on the generated binary codes.”
Unsupervised Semantic-Preserving Adversarial Hashing for Image Search (2019, background): “Other recent deep hashing methods learn hash codes by maximizing their representation capacity [46] or enforcing the similarity between rotated images and their corresponding original images [47].”
Two-Stream Deep Hashing With Class-Specific Centers for Supervised Image Search (2020, background): “DH [51] develops a neural network to seek multiple hierarchical nonlinear transformations to map data points to Hamming space.”
A Survey on Learning to Hash (2016, methodology): “Following the scalar quantization approach, deep hashing [80] defines a loss to penalize the difference between the binary hash codes (see Equation (35)) and the real values fromwhich a linear projection is used to generate the binary codes, and”
Learning Compact Binary Descriptors with Unsupervised Deep Neural Networks (2016, background): “Deep Hashing (DH) [24] builds three layers hierarchical neural networks to learn discriminative projection matrix, but their method does not take the advantage of deep transfer learning, thus makes the binary codes less effective.”
Deep Cauchy Hashing for Hamming Space Retrieval (2018, background): “Recently, deep learning to hash methods [33, 17, 28, 8, 36, 1, 21, 4] yield breakthrough results on image retrieval datasets by blending the power of deep learning [15, 13].”