Zero-Shot Learning (ZSL) aims at recognizing unseen classes that are absent during the training stage. Unlike the existing approaches that learn a visual-semantic embedding model to bridge the ...low-level visual space and the high-level class prototype space, we propose a novel synthesized approach for addressing ZSL within a dictionary learning framework. Specifically, it learns both a dictionary matrix and a class-specific encoding matrix for each seen class to synthesize pseudo instances for unseen classes with auxiliary of seen class prototypes. This allows us to train the classifiers for the unseen classes with these pseudo instances. In this way, ZSL can be treated as a traditional classification task, which makes it applicable for traditional and generalized ZSL settings simultaneously. Extensive experimental results on four benchmark datasets (AwA, CUB, aPY, and SUN) demonstrate that our method yields competitive performances compared to state-of-the-art methods on both settings.
In this paper, we propose a remote sensing image fusion method which combines the wavelet transform and sparse representation to obtain fusion images with high spectral resolution and high spatial ...resolution. Firstly, intensity-hue-saturation (IHS) transform is applied to Multi-Spectral (MS) images. Then, wavelet transform is used to the intensity component of MS images and the Panchromatic (Pan) image to construct the multi-scale representation respectively. With the multi-scale representation, different fusion strategies are taken on the low-frequency and the high-frequency sub-images. Sparse representation with training dictionary is introduced into the low-frequency sub-image fusion. The fusion rule for the sparse representation coefficients of the low-frequency sub-images is defined by the spatial frequency maximum. For high-frequency sub-images with prolific detail information, the fusion rule is established by the images information fusion measurement indicator. Finally, the fused results are obtained through inverse wavelet transform and inverse IHS transform. The wavelet transform has the ability to extract the spectral information and the global spatial details from the original pairwise images, while sparse representation can extract the local structures of images effectively. Therefore, our proposed fusion method can well preserve the spectral information and the spatial detail information of the original images. The experimental results on the remote sensing images have demonstrated that our proposed method could well maintain the spectral characteristics of fusion images with a high spatial resolution.
Cloud-assisted Internet of Things (IoT) overcomes the resource-constrained nature of the traditional IoT and is developing rapidly in such fields as smart grids and intelligent transportation. In a ...cloud-assisted IoT system, users can remotely control the IoT devices and send specific instructions to them. If the users' identities are not verified, adversaries can pretend as legitimate users to send fake and malicious instructions to IoT devices, thereby compromising the security of the entire system. Thus, a sound authentication mechanism is indispensable to ensure security. At the same time, it should be noted that a gateway may connect to massive IoT devices with the exponential growth of interconnected devices in a cloud-assisted IoT system. The efficiency of authentication schemes is easily impacted by the computation capability of the gateway. Recently, several schemes have been designed for cloud-assisted IoT systems, but they have problems of one kind or another, making them not very suitable for cloud-assisted IoT systems. In this paper, we take a typical scheme (proposed at IEEE TDSC 2020) as an example to identify the common weaknesses and challenges of designing a user authentication scheme for cloud-assisted IoT systems. In addition, we propose a new secure user authentication scheme with lightweight computation on gateways. The proposed scheme provides secure access between remote users and IoT devices with many ideal attributions, such as forward secrecy and multi-factor security. Meanwhile, the security of this scheme is proved under the random-oracle model, heuristic analysis, the ProVerif tool and BAN logic. Compared with ten state-of-the-art schemes in security and performance, the proposed scheme achieves all the listed twelve security requirements with minimum computation and storage costs on gateways.
Collaborative filtering (CF) is a widely applied method to perform recommendation tasks in a wide range of domains and applications. Dictionary learning (DL) models, which are highly important in ...CF-based recommender systems (RSs), are well represented by rating matrices. However, these methods alone do not resolve the cold start and data sparsity issues in RSs. We observed a significant improvement in rating results by adding trust information on the social network. For that purpose, we proposed a new dictionary learning technique based on trust information, called TrustDL, where the social network data were employed in the process of recommendation based on structural details on the trusted network. TrustDL sought to integrate the sources of information, including trust statements and ratings, into the recommendation model to mitigate both problems of cold start and data sparsity. It conducted dictionary learning and trust embedding simultaneously to predict unknown rating values. In this paper, the dictionary learning technique was integrated into rating learning, along with the trust consistency regularization term designed to offer a more accurate understanding of the feature representation. Moreover, partially identical trust embedding was developed, where users with similar rating sets could cluster together, and those with similar rating sets could be represented collaboratively. The proposed strategy appears significantly beneficial based on experiments conducted on four frequently used datasets: Epinions, Ciao, FilmTrust, and Flixster.
•Blending sources alignment (BSA) is designed to unify multiple domains.•Feature-oriented unified dictionary learning (FUDL) framework is developed to learn a discriminative feature-oriented ...dictionary in the unified domain.•Feature-oriented unified dictionary (FUD) to sparsely reconstruct the optimally weighted feature.•FUDL-SC has been validated in fulfilling the multi-domain fault diagnosis tasks.•It is the first attempt to combine sample-domain adaption and feature-domain generalization outside of the deep learning framework.
Aiming at the instability of sparse representation of multi-domain features caused by domain discrepancy and individual differences in feature sparse discriminative ability, the Feature-oriented Unified Dictionary Learning based Sparse Classification (FUDL-SC) is proposed for multi-domain fault diagnosis. Blending sources alignment is designed to unify the transformations of multiple domains at the sample global level to reduce the discrepancy in the data distribution across domains. The FUDL framework is established by learning dictionary atoms specific to each feature of different classes in the unified domain, thus generating a Feature-oriented Unified Dictionary (FUD). Reconstruction Scoring Matrix (RSM) is constructed to measure the sparse discrimination performance of individual FUD. The iterative update of the RSM is incorporated into the FUDL model, thus enhancing FUD's sparsity-preserving and discriminating ability in multiple domains. The efficacy of FUDL-SC is experimentally demonstrated on a multi-condition bearing failure dataset, a multi-device bearing failure dataset, and a small-sample gearbox failure dataset. The comparative studies verify that FUDL-SC can achieve higher recognition accuracy, better stability, and greater robustness than numerous state-of-the-art methods in multi-domain fault diagnosis.
•Demonstrate use of BrainMap or Neurosynth as a cognition dictionary.•Less than 1% of brain imaging studies have utilized this dictionary approach.•This approach can be especially valuable for ...clinical research and practice.
Characterizing the functional involvement of specific brain regions has long been a central challenge in cognitive neuroscience. Functional magnetic resonance imaging (fMRI) techniques have offered solutions for mapping functional neural networks. The complex nature of structure-function correspondence makes an elaborate task design difficult to fully capture higher-order cognitive function. Other research practices, such as brain-behavior association or between-group comparisons, are thus widely used to explore cognitive correlations with specific brain regions. However, interpreting the results derived from a specific brain region with their underlying cognitive functions has been too general in publications. Here, we use two examples, i.e., a brain-intelligence correlation study and a depression-control comparison meta-study, to demonstrate use of two neuroimaging online databases, BrainMap and Neurosynth. One key utility of the two databases is collecting results from massive cognitive task-based fMRI (tb-fMRI) studies, i.e., coordinates in standard brain space. Just like looking up a “coordinate-based cognition dictionary”, researchers can receive a plethora of related tb-fMRI activation information characterized by cognitive domains, specific cognitive functions, cognitive task paradigms, and related publications. Surprisingly, we found that only less than 1% of brain-behavior association or between-group comparison studies have utilized this dictionary approach. We encourage the community to further engage with the existing databases for specific and comprehensive interpretation of neuroimaging as well as guidance of future experimental tb-fMRI design.
Planet bearing fault identification is an attractive but challenging task in wind turbine condition monitoring and fault diagnosis. Traditional fault characteristic frequency identification based ...diagnostic strategies are not sufficient for reliable planet bearing fault detection, due to complex physical configurations and modulation characteristics in planetary drivetrains. In this paper, we propose a discriminative dictionary learning based sparse representation classification (DDL-SRC) framework for intelligent planet bearing fault identification. Our framework could learn a reconstructive and discriminative dictionary for signal sparse representation and an optimal linear classifier for classification tasks simultaneously, which bridges the gap between dictionary learning and classifier training in traditional SRC methods. Specifically, the optimization objective introduces a novel regularization term called ‘discriminative sparse codes error’ and incorporates it with the reconstruction error and classification error. Thus, the dictionary learned by our framework possesses not only the reconstructive power for sparse representation but also the discriminative power for classifier training. The optimization formulation is efficiently solved using K-SVD and orthogonal matching pursuit algorithms. The experiment validations have been conducted for demonstrating the effectiveness and superiority of the proposed DDL-SRC framework over the state-of-the-art dictionary learning based SRC and deep convolutional neural network methods for intelligent planet bearing fault identification.
•A new discriminative dictionary learning based sparse representation classification method is proposed for fault diagnosis.•DDL-SRC introduces a discriminative sparse codes error term to enhance the discriminability for dictionary learning.•DDL-SRC learns a discriminative dictionary and an optimal classifier model jointly for superior classification performance.•The optimization problem for discriminative dictionary learning is efficiently solved by the K-SVD algorithm.•DDL-SRC outperforms the state-of-the-art methods for intelligent fault identification of planet bearings in wind turbine.
In this paper, a local adaptive joint sparse representation (LAJSR) model is proposed for the classification of hyperspectral remote sensing images. It improves the original joint sparse ...representation (JSR) method in both the signal and dictionary construction phase and sparse representation phase. Given a testing pixel, a similar signal set is constructed by picking a few of the most similar pixels from its spatial neighborhood. The original training dictionary consists of training samples from different classes and is extended by adding spatial neighbors of each training sample. A local adaptive dictionary is built by selecting the most representative atoms from the extended dictionary that are correlated to the similar signal set. In the LAJSR framework, the selected similar signals are simultaneously represented by the local adaptive dictionary, and the obtained sparse representation coefficients are further weighted by a sparsity concentration index vector which aims to concentrate and highlight the coefficients on the expected class. Experimental results on two benchmark hyperspectral data sets have demonstrated that the proposed LAJSR method is much more effective than existing JSR and SVM methods, especially in the case of small sample sizes.
In this article we present a novel dictionary learning framework designed for compression and sampling of light fields and light field videos. Unlike previous methods, where a single dictionary with ...one-dimensional atoms is learned, we propose to train a Multidimensional Dictionary Ensemble (MDE). It is shown that learning an ensemble in the native dimensionality of the data promotes sparsity, hence increasing the compression ratio and sampling efficiency. To make maximum use of correlations within the light field data sets, we also introduce a novel nonlocal pre-clustering approach that constructs an Aggregate MDE (AMDE). The pre-clustering not only improves the image quality but also reduces the training time by an order of magnitude in most cases. The decoding algorithm supports efficient local reconstruction of the compressed data, which enables efficient real-time playback of high-resolution light field videos. Moreover, we discuss the application of AMDE for compressed sensing. A theoretical analysis is presented that indicates the required conditions for exact recovery of point-sampled light fields that are sparse under AMDE. The analysis provides guidelines for designing efficient compressive light field cameras. We use various synthetic and natural light field and light field video data sets to demonstrate the utility of our approach in comparison with the state-of-the-art learning-based dictionaries, as well as established analytical dictionaries.