Topic
Feature hashing
About: Feature hashing is a research topic. Over the lifetime, 993 publications have been published within this topic receiving 51462 citations.
Papers published on a yearly basis
Papers
More filters
••
TL;DR: Wang et al. as discussed by the authors proposed a novel hashing method named multi-manifold deep discriminative cross-modal hashing (MDDCH) for large-scale medical image retrieval.
Abstract: Benefitting from the low storage cost and high retrieval efficiency, hash learning has become a widely used retrieval technology to approximate nearest neighbors. Within it, the cross-modal medical hashing has attracted an increasing attention in facilitating efficiently clinical decision. However, there are still two main challenges in weak multi-manifold structure perseveration across multiple modalities and weak discriminability of hash code. Specifically, existing cross-modal hashing methods focus on pairwise relations within two modalities, and ignore underlying multi-manifold structures across over 2 modalities. Then, there is little consideration about discriminability, i.e., any pair of hash codes should be different. In this paper, we propose a novel hashing method named multi-manifold deep discriminative cross-modal hashing (MDDCH) for large-scale medical image retrieval. The key point is multi-modal manifold similarity which integrates multiple sub-manifolds defined on heterogeneous data to preserve correlation among instances, and it can be measured by three-step connection on corresponding hetero-manifold. Then, we propose discriminative item to make each hash code encoded by hash functions be different, which improves discriminative performance of hash code. Besides, we introduce Gaussian-binary Restricted Boltzmann Machine to directly output hash codes without using any continuous relaxation. Experiments on three benchmark datasets (AIBL, Brain and SPLP) show that our proposed MDDCH achieves comparative performance to recent state-of-the-art hashing methods. Additionally, diagnostic evaluation from professional physicians shows that all the retrieved medical images describe the same object and illness as the queried image.
4 citations
••
01 Jul 2018TL;DR: Experimental results on prediction tasks with hundred-millions of features demonstrate that CCFH can achieve the same level of performance by using only 15%-25% parameters compared with conventional feature hashing.
Abstract: Feature hashing is widely used to process large scale sparse features for learning of predictive models. Collisions inherently happen in the hashing process and hurt the model performance. In this paper, we develop a feature hashing scheme called Cuckoo Feature Hashing(CCFH) based on the principle behind Cuckoo hashing, a hashing scheme designed to resolve collisions. By providing multiple possible hash locations for each feature, CCFH prevents the collisions between predictive features by dynamically hashing them into alternative locations during model training. Experimental results on prediction tasks with hundred-millions of features demonstrate that CCFH can achieve the same level of performance by using only 15%-25% parameters compared with conventional feature hashing.
4 citations
•
TL;DR: A new image-hashing algorithm using Harris corners and singular value decomposition is proposed, which is stable to visually insignificant changes due to normal image processing and JPEG coding, while sensitive to excessive changes and malicious tampering.
Abstract: Perceptual image hashing maps an image to a short data string,applicable to image authentication,content-based image retrieval,digital watermarking,etc.We propose a new image-hashing algorithm using Harris corners and singular value decomposition.Critical feature points robust against gray-level modification and image rotation are identified.A prescribed number of large singular values of the image blocks centered at the robust feature points are quantized to compress the data,which represent positions of the points and information of their neighborhood.The compressed data are then coded to generate the hash.The obtained hash is stable to visually insignificant changes due to normal image processing and JPEG coding,while sensitive to excessive changes and malicious tampering.Security of the hash is guaranteed by using secret keys.
4 citations
••
TL;DR: This is the first work that introduces hash-based similarity search method to perform fingerprint detection and demonstrates that the proposed approach outperforms traditional linear scan detection methods in term of efficiency.
Abstract: Digital fingerprinting is a promising approach to protect multimedia contents from unauthorized redistribution. Whereas, large scale and high dimensionality make existing fingerprint detection methods fail to trace the traitors efficiently. To handle this problem, we propose a novel local and global structure preserving hashing to conduct fast fingerprint detection. This is the first work that introduces hash-based similarity search method to perform fingerprint detection. Applying the hashing method, we obtain a neighborhood-preserving low-dimensional representation (e. g. hash code) for each fingerprint. Through hash codes, we can find the nearest neighbors of the extracted fingerprint, thereby tracing the real traitors within a small range. Preserving the local structure facilitates to find the nearest neighbors of the query fingerprint efficiently, and preserving the global structure ensures hash codes of fingerprints as discriminative as possible. These properties make the proposed approach efficient to trace the real traitors. Extensive experiments demonstrate that the proposed approach outperforms traditional linear scan detection methods in term of efficiency.
4 citations
••
TL;DR: This paper finds for the first time that the default vector size of current feature hashing practices is unnecessarily large, which reduces memory space by 70% and increases the detection accuracy, compared with the state-of-the-art scheme.
4 citations