To improve the effectiveness and efficiency of biomedical information retrieval by proposing ranking-based methods for selecting an optimal subset of retrieval systems for data fusion, we propose three ranking-based subset selection methods SFS (Sequential Forward Search), D&P (Diversity & Performance), and P&D (Performance & Diversity). These methods were applied in combination with the Reciprocal Rank Fusion technique. Experiments were conducted on four medical datasets from TREC, using between 62 and 125 candidate retrieval systems, and selecting up to 15 for fusion. The proposed subset selection methods significantly improved retrieval performance. Fusing the selected systems using RRF yielded improvements ranging from 10% to over 60% compared to the best individual retrieval system across the datasets. They also outperform the state-of-the-art technology by a large margin. In summary, our subset selection approach offers a practical and cost-efficient solution for biomedical information retrieval, achieving substantial performance gains while reducing computational overhead.

In multi-label learning, traditional methods try to directly establish mapping functions between samples and their labels. However, such methods may suffer low classification performance due to inherent noise and incoherent representation of samples. Therefore, considering the correlation between samples and their neighbors, as they may share common feature semantics, a multi-label classification method via feature enhancement from neighborhood samples, referred to as DRNSA is proposed. In this method, we construct a Laplacian graph dynamically, by considering the distance of two samples in a projected low rank subspace. With this neighborhood selection strategy, we then build a multi-label classifier via feature enhancement from neighborhood samples. Different from past works that directly build classifiers from samples and their labels, we build a new enhanced data sample which is weighted by its semantically similar neighborhood samples. Thus, our method can obtain better robust subspace in very high noisy data representations. Experiments conducted on cal500, corel5k, corel16k1 and corel16k4 datasets, show a significant out-performance of our DRNSA method over multi-label classification methods, including MLSF, MLFE, BR, PLST, CSSP, CPLST, and FaIE.

Next Basket Recommender systems in e-commerce face challenges such as data sparsity, evolving user preferences, and cross-domain transfer limitations. We propose the Temporal Dual-Target Cross-Domain Recommendation Framework (T-DualCRF) to address these issues. T-DualCRF integrates multi-channel embeddings (user feedback, knowledge graphs, temporal features) and a dual-target mechanism for robust cross-domain knowledge transfer. It also employs time-aware embeddings and a temporal heterogeneous graph to model user preference changes. The framework’s hybrid optimization mechanism, combining the Multi-Verse Optimizer and Whale Optimization Algorithm, enhances recommendation accuracy and stability. Experimental results on Amazon datasets show that T-DualCRF significantly outperforms existing models, with improvements of up to 20% in F1-score and 17% in NDCG, effectively mitigating data sparsity and adapting to real-time user behavior changes.

Canonical correlation analysis (CCA) is a correlation analysis technique that is widely used in statistics and the machine-learning community. However, the high complexity involved in the training process lays a heavy burden on the processing units and memory system, making CCA nearly impractical in large-scale data. To overcome this issue, a novel CCA method that tries to carry out analysis on the dataset in the Fourier domain is developed in this article. Appling Fourier transform on the data, we can convert the traditional eigenvector computation of CCA into finding some predefined discriminative Fourier bases that can be learned with only element-wise dot product and sum operations, without complex time-consuming calculations. As the eigenvalues come from the sum of individual sample products, they can be estimated in parallel. Besides, thanks to the data characteristic of pattern repeatability, the eigenvalues can be well estimated with partial samples. Accordingly, a progressive estimate scheme is proposed, in which the eigenvalues are estimated through feeding data batch by batch until the eigenvalues sequence is stable in order. As a result, the proposed method shows its characteristics of extraordinarily fast and memory efficiencies. Furthermore, we extend this idea to the nonlinear kernel and deep models and obtained satisfactory accuracy and extremely fast training time consumption as expected. An extensive discussion on the fast Fourier transform (FFT)-CCA is made in terms of time and memory efficiencies. Experimental results on several large-scale correlation datasets, such as MNIST8M, X-RAY MICROBEAM SPEECH, and Twitter Users Data, demonstrate the superiority of the proposed algorithm over state-of-the-art (SOTA) large-scale CCA methods, as our proposed method achieves almost same accuracy with the training time of our proposed method being 1000 times faster. This makes our proposed models best practice models for dealing with large-scale correlation datasets. The source code is available at https://github.com/Mrxuzhao/FFTCCA.

Traditional low-rank methods overlook residuals as corruptions, but we discovered that low-rank residuals actually keep image edges together with corrupt components. Therefore, filtering out such structural information could hamper the discriminative details in images, especially in heavy corruptions. In order to address this limitation, this paper proposes a novel method named ESL-LRR, which preserves image edges by finding image projections from low-rank residuals. Specifically, our approach is built in a manifold learning framework where residuals are regarded as another view of image data. Edge preserved image projections are then pursued using a dynamic affinity graph regularization to capture the more accurate similarity between residuals while suppressing the influence of corrupt ones. With this adaptive approach, the proposed method can also find image intrinsic low-rank representation, and much discriminative edge preserved projections. As a result, a new classification strategy is introduced, aligning both modalities to enhance accuracy. Experiments are conducted on several benchmark image datasets, including MNIST, LFW, and COIL100. The results show that the proposed method has clear advantages over compared state-of-the-art (SOTA) methods, such as Low-Rank Embedding (LRE), Low-Rank Preserving Projection via Graph Regularized Reconstruction (LRPP_GRR), and Feature Selective Projection (FSP) with more than 2% improvement, particularly in corrupted cases.

Synthetic aperture radar (SAR) image despeckling is a challenging task as speckle noise is spatially correlated and signal-dependent, and appears as a grainy texture superimposed on images. Although traditional low-rank SAR image despeckling methods have shown promising performance, they have the problem of producing over-smoothed images with blurred edges due to their low-rank characteristics. In this paper, we propose a novel edge preserved SAR despeckling method named EP-LRSID, which can keep rich edge details while reducing speckle noise. Specifically, EP-LRSID takes a fresh look at the low-rank model, i.e., we can obtain structural edge information from residuals which is viewed as noise and simply disregarded by the traditional low-rank methods. To obtain discriminative edge information from residuals, the edge subspace is obtained in a manifold framework by using a dynamic affinity graph regularization. Moreover, a new hierarchical prior knowledge regulation is designed to make different kinds of pixels processed hierarchically, especially the strong scattering points in SAR images. By introducing this prior knowledge, our low-rank model can obtain more confidential low-rank parts and edge parts, thus structural information including edges can be better preserved in this way. Extensive experiments on several real and synthetic datasets demonstrate that EP-LRSID can achieve the highest despeckling performance with edge preservation than other state-of-the-art despeckling algorithms.

Manifold learning is a widely used technique for dimensionality reduction as it can reveal the intrinsic geometric structure of data. However, its performance decreases drastically when data samples are contaminated by heavy noise or occlusions, which leads to unsatisfying data processing performance. We propose a novel robust dimensionality reduction method via low-rank Laplacian graph learning for classification and clustering tasks to solve the above problem. First, we construct a low-rank Laplacian graph by combining manifold learning and subspace learning. This graph can capture both global and local structural information of the data. And we introduce rank constraints for the Laplacian graph to make it more discriminative. Second, we put the learning of projection matrix and sample affinity graph into a unified framework. The projection matrix is embedded into a robust low-rank Laplacian graph so that the low-dimensional mapping of data can maintain the structural information in the graph well. Finally, we add a regularization term to the projection matrix to make it have the ability of both feature extraction and feature selection. Therefore, the proposed model can resist the interference of noise or data damage to learn the optimal projection to achieve better performance in dimensionality reduction through such a data dimensionality reduction joint framework. Comprehensive experiments on various benchmark datasets with varying degrees of occlusions or corruptions are carried out to evaluate the performance of the proposed method. Compared with the state-of-the-art dimensionality reduction methods in the literature, the experimental results are inspiring, showing our method’s effectiveness and robustness in classification and clustering, especially in object recognition scenarios with noise or occlusions.

Multi-label classification remains a challenging task for high-dimensional data samples and their labels both increase the complexity of training models. In this paper, we propose a Robust Label and Feature Space Co-Learning method, referred to as RLFSCL, for multi-label classification. Different from traditional multi-label classification methods which focus on feature space learning through regression directly between data samples and labels, our proposed method can further learn robust low rank label space from this traditional regression method. Therefore, our RLFSCL can learn better low rank feature and label representations simultaneously in original noisy and high dimensional spaces. Experimental comparison on five benchmark datasets, including Rcv1s5, Cal500, and Corel16k4 shows that the proposed RLFSCL algorithm outperforms state-of-the-art multi-label classification methods. The code of RLFSCL is made available onhttps://github.com/JingChuanTang/RLFSCL.

Cross domain adaptation aims to improve the performance of the target domain model by making full use of information rich source domain samples. However, as information becomes richer, the noise also increases. In order to improve the reliability of cross domain adaptation, we propose a novel method based on deep robust low rank correlation. Borrowed from the traditional idea of Canonical Correlation Analysis (CCA), we developed a robust correlation model to maximize the correlation between source and target domains. Also, the low-rank characteristics of cross domain data can effectively reduce the negative influence of noisy data. Furthermore, in order that the cross-domain data can share a unifying clustering structure, we introduced a common Laplacian affinity structure. Then the learned features can be smoothed and aligned to the unifying structure. In this way, we obtain a deep robust low rank correlation model with the help of the unifying clustering structure, which can effectively reduce the influence of noise and improve the performance of cross domain adaptation. Experimental results on three datasets including Office-31, ImageCLEF-DA and Office-Home show that our model significantly outperforms state-of-the-art cross domain adaptation methods.

Smoothing images while preserving salient edges is a crucial task in computational photography. Existing edge-preserving filters suffer from various artifacts, such as halos, gradient reversals, and intensity shifts. Observing that various artifacts are strongly related to salient edges with large gradients, we propose a continuous mapping function to process the gradients. The proposed function is literally edge-preserving, i.e., it keeps large gradients intact while attenuating small gradients. We propose an L1-regularized reconstruction model based on the processed gradients for edge-preserving image filtering. The L1-regularization facilitates the edge-preserving property in the reconstructed results. To solve the proposed L1-regularized model, we implement an efficient algorithm based on the alternating direction method of multipliers (ADMM) and Fourier domain optimization. We have conducted qualitative and quantitative experiments to evaluate the proposed filter. The results demonstrate that our filter better handles various artifacts and delivers superior image quality on various applications. The proposed filter is highly efficient, our GPU implementation takes 70ms to process a color image with 1 megapixel on an NVIDIA GTX 1070 GPU.

Canonical correlation analysis (CCA) is a linear correlation analysis technique used widely in the statistics and machine learning community. However, the high complexity involved in pursuing eigenvector lays a heavy burden on the memory and computational time, making CCA nearly impractical in large-scale cases. In this paper, we attempt to overcome this issue by representing the data in the Fourier domain. Thanks to the data characteristic of pattern repeatability, one can translate projection-seeking of CCA into choosing some discriminative Fourier bases with only element-wise dot product and sum operations, without time-consuming eigenvector computation. Another merit of this scheme is that the eigenvalues can be approximated asymptotically in contrast to existing methods. Specifically, the eigenvalues can be estimated progressively, and the accuracy goes up as the number of data samples increases monotonously. This makes it possible to use partial data samples to obtain satisfactory accuracy. All the facts above make the proposed method extremely fast and memory efficient. Experimental results on several large-scale datasets, such as MNIST 8M, X-RAY MICROBEAM SPEECH, and TWITTER USERS Data, demonstrate the superiority of the proposed algorithm over SOTA large-scale CCA methods, as our proposed method achieves almost same accuracy with the training time being 1,000 times faster than SOTA methods.

Depth-of-field (DoF) rendering is an important technique in computational photography that simulates the human visual attention system. Existing DoF rendering methods usually suffer from a high computational cost. The task of DoF rendering can be accelerated by guided upsampling methods. However, the state-of-the-art guided upsampling methods fail to distinguish the focus and defocus areas, resulting in unsatisfying DoF effects. In this paper, we propose a novel deep weighted guided upsampling network (DWGUN) based on a encoder and decoder framework to jointly upsample the low-resolution DoF image under the guidance of the corresponding high-resolution all-in-focus image. Due to the intuitive weight design, the traditional weighted image upsampling is not tailored to DoF image upsampling. We propose a deep refocus-defocus edge-aware module (DREAM) to learn the spatially-varying weights and embed them in the deep weighted guided upsampling block (DWGUB). We have conducted comprehensive experiments to evaluate the proposed method. Rigorous ablation studies are also conducted to validate the rationality of the proposed components.

Abstract In the last years, subspace‐based multi‐view face recognition has attracted increasing attention and many related methods have been proposed. However, the most existing methods ignore the specific local structure of different views. This drawback can cause these methods' discriminating ability to degrade when many noisy samples exist in data. To tackle this problem, a multi‐view low‐rank representation method is proposed, which exploits both intrinsic relationships and specific local structures of different views simultaneously. It is achieved by hierarchical Bayesian methods that constrain the low‐rank representation of each view so that it matches a linear combination of an intrinsic representation matrix and a specific representation matrix to obtain common and specific characteristics of different views. The intrinsic representation matrix holds the consensus information between views, and the specific representation matrices indicate the diversity among views. Furthermore, the model injects a clustering structure into the low‐rank representation. This approach allows for adaptive adjustment of the clustering structure while pursuing the optimization of the low‐rank representation. Hence, the model can well capture both the relationship between data and the clustering structure explicitly. Extensive experiments on several datasets demonstrated the effectiveness of the proposed method compared to similar state‐of‐the‐art methods in classification and clustering.

Multi-view subspace clustering (MVSC) finds a shared structure in latent low-dimensional subspaces of multi-view data to enhance clustering performance. Nonetheless, we observe that most existing MVSC methods neglect the diversity in multi-view data by considering only the common knowledge to find a shared structure either directly or by merging different similarity matrices learned for each view. In the presence of noise, this predefined shared structure becomes a biased representation of the different views. Thus, in this article, we propose a MVSC method based on coupled low-rank representation to address the above limitation. Our method first obtains a low-rank representation for each view, constrained to be a linear combination of the view-specific representation and the shared representation by simultaneously encouraging the sparsity of view-specific one. Then, it uses the k -block diagonal regularizer to learn a manifold recovery matrix for each view through respective low-rank matrices to recover more manifold structures from them. In this way, the proposed method can find an ideal similarity matrix by approximating clustering projection matrices obtained from the recovery structures. Hence, this similarity matrix denotes our clustering structure with exactly k connected components by applying a rank constraint on the similarity matrix’s relaxed Laplacian matrix to avoid spectral post-processing of the low-dimensional embedding matrix. The core of our idea is such that we introduce dynamic approximation into the low-rank representation to allow the clustering structure and the shared representation to guide each other to learn cleaner low-rank matrices that would lead to a better clustering structure. Therefore, our approach is notably different from existing methods in which the local manifold structure of data is captured in advance. Extensive experiments on six benchmark datasets show that our method outperforms 10 similar state-of-the-art compared methods in six evaluation metrics.

In this article, a novel ensemble model, called Multiple Kernel Ensemble Learning (MKEL), is developed by introducing a unified ensemble loss. Different from the previous multiple kernel learning (MKL) methods, which attempt to seek a linear combination of basis kernels as a unified kernel, our MKEL model aims to find multiple solutions in corresponding Reproducing Kernel Hilbert Spaces (RKHSs) simultaneously. To achieve this goal, multiple individual kernel losses are integrated into a unified ensemble loss. Therefore, each model can co-optimize to learn its optimal parameters by minimizing a unified ensemble loss in multiple RKHSs. Furthermore, we apply our proposed ensemble loss into the deep network paradigm and take the sub-network as a kernel mapping from the original input space into a feature space, named Deep-MKEL (D-MKEL). Our D-MKEL model can utilize the diversified deep individual sub-networks into a whole unified network to improve the classification performance. With this unified loss design, our D-MKEL model can make our network much wider than other traditional deep kernel networks and more parameters are learned and optimized. Experimental results on several mediate UCI classification and computer vision datasets demonstrate that our MKEL model can achieve the best classification performance among comparative MKL methods, such as Simple MKL, GMKL, Spicy MKL, and Matrix-Regularized MKL. On the contrary, experimental results on large-scale CIFAR-10 and SVHN datasets concretely show the advantages and potentialities of the proposed D-MKEL approach compared to state-of-the-art deep kernel methods.

Cross-domain data has become very popular recently since various viewpoints and different sensors tend to facilitate better data representation. In this article, we propose a novel cross-domain object representation algorithm (RLRCA) which not only explores the complexity of multiple relationships of variables by canonical correlation analysis (CCA) but also uses a low rank model to decrease the effect of noisy data. To the best of our knowledge, this is the first try to smoothly integrate CCA and a low-rank model to uncover correlated components across different domains and to suppress the effect of noisy or corrupted data. In order to improve the flexibility of the algorithm to address various cross-domain object representation problems, two instantiation methods of RLRCA are proposed from feature and sample space, respectively. In this way, a better cross-domain object representation can be achieved through effectively learning the intrinsic CCA features and taking full advantage of cross-domain object alignment information while pursuing low rank representations. Extensive experimental results on CMU PIE, Office-Caltech, Pascal VOC 2007, and NUS-WIDE-Object datasets, demonstrate that our designed models have superior performance over several state-of-the-art cross-domain low rank methods in image clustering and classification tasks with various corruption levels.

In this paper, a novel manifold alignment approach via multi-graph embedding (MA-MGE) is proposed. Different from the traditional manifold alignment algorithms that use a single graph to describe the latent manifold structure of each dataset, our approach utilizes multiple graphs for modeling multiple local manifolds in multi-view data alignment. Therefore a composite manifold representation with complete and more useful information is obtained from each dataset through a dynamic reconstruction of multiple graphs. Experimental results on Protein and Face-10 datasets demonstrate that the mapping coordinates of the proposed method provide better alignment performance compared to the state-of-the-art methods, such as semi-supervised manifold alignment (SS-MA), manifold alignment using Procrustes analysis (PAMA) and manifold alignment without correspondence (UNMA).

This study proposes a novel dimensionality reduction (DR) method for multi‐view datasets. The principal component analysis (PCA) idea of minimising least squares reconstruction errors is extended to consider both data distribution and penalty weights called dictionary to recover outliers free global structures from missing and noisy data points. In this way, PCA is viewed as a special instance of the authors’ proposed dictionary induced least squares framework (DLS). Furthermore, to appropriately handle multi‐view DR, we combine the DLS with multiple manifold embeddings (DLSME). Therefore it can obtain lower projections while maintaining a balance between preserving global structures with DLS and local structures with multi‐manifold embeddings. Extensive experiments on object and face recognition datasets verify that the DLS achieves better classification results with lower dimensional projections than PCA. Also, on many multi‐view datasets of visual recognition and web image annotation, the DLSME method demonstrates more effectiveness than Graph‐Laplacian PCA (gLPCA), robust PCA‐optimal mean, canonical correlation analysis (CCA), bilinear models (BLM), neighbourhood preserving embedding, locality preserving projections, and locality sensitive discriminant analysis.

Kernel canonical correlation analysis (KCCA) is an efficient dimensionality reduction tool in the application of remote sensing image classification. However, it suffers from the problem of parametric sensitivity since a single kernel is used. In this letter, a KCCA ensemble framework is put forward to improve the robustness of KCCA. Following the philosophy that two heads are better than one, multiple KCCA models are incorporated into the framework. And more importantly, their terms are weighted to adjust their contribution to the result according to their performance. In addition, over-fitting is overcome by introducing a Laplacian regularization term in our framework, hence, the name Laplacian regularized kernel canonical correlation ensemble. Experimental results on NWPU-RESISC45 data set show that our proposed method achieves better classification performances as compared to state-of-the-art methods in both shallow and deep features.

Feature learning has enjoyed much attention and achieved good performance in recent studies of image processing. Unlike the required training conditions often assumed there, far less labeled data is available for training emotion classification systems. In addition, current feature learning is typically performed on an entire face image without considering the dependency between features. These approaches ignore the fact that faces are structured and the neighboring features are dependent. Thus, the learned features lack the power to describe visually coherent facial images. Our method is therefore designed with the goal of simplifying the problem domain by removing expression-irrelevant factors from the input images, with a key region-based mechanism, which is an effort to reduce the amount of data required to effectively train the feature-learning methods. Meanwhile, we can construct geometric constraints between the key regions and its detected positions. To this end, we introduce a Spatially Coherent featurelearning method for Pose-invariant Facial Expression Recognition (SC-PFER). In our model, we first perform face frontalization through a 3D pose-normalization technique, which could normalize poses while preserving the identity information through synthesizing frontal faces for facial images with arbitrary views. Subsequently, we select a sequence of key regions around 51 key points in the synthetic frontal face images for efficient unsupervised feature learning. Finally, we introduce a linkage structure over the learning-based features and the corresponding geometry information of each key region to encode the dependencies of the regions. Our method, on the whole, does not require training multiple models for each specific pose and avoids separating training and parameter tuning for each pose. The proposed framework has been evaluated on two benchmark databases, BU-3DFE and SFEW, for pose-invariant Facial Expression Recognition (FER). The experimental results demonstrate that our algorithm outperforms current state-of-the-art FER methods. Specifically, our model achieves an improvement of 1.72% and 1.11% FER accuracy, on average, on BU-3DFE and SFEW, respectively.

Ensemble Learning is widely accepted as an effective technique to improve accuracy and stability of a single classifier. Classifier ensemble generally should combine diverse component classifiers. Accuracy and diversity are two key factors to decide the ensemble generalization error. In this paper, we propose a novel approach to make the tradeoff between accuracy and diversity by maximizing accuracy and diversity simultaneously in an ensemble classifier. Our proposed method is a minimization convex optimization problem. Experimental results on a variety of UCI and artificial datasets have shown that, our proposed method has advantage of superior performances in keeping higher classification results than other ensemble methods, such as Random Forest, AdaBoost, EnsembleSVM and Weighted Classifier Ensemble method Based on Quadratic Forms(QFWEC).

In this paper, we propose a multi-local means based nearest neighbor classifier (MLMNN). In the MLMNN, k categorical nearest neighbors of a query sample are first found and used to calculate the corresponding k categorical multi-local mean vectors which can represent different local class-specific sample distributions. Then, the query sample is represented by a linear combination of k categorical local mean vectors and the representation coefficient of each local mean vector as the contribution to representing and classifying the query sample is obtained. Finally, the class-specific representation-based distance (i.e. reconstruction residual) between the query sample and k categorical multi-local mean vectors is adopted to determine the class label of the query sample. The experimental results on three popular face databases show that the proposed MLMNN method outperforms the related competitive KNN-based methods.

Emotion recognition via facial expressions (ERFE) has attracted a great deal of interest with recent advances in artificial intelligence and pattern recognition. Most studies are based on 2D images, and their performance is usually computationally expensive. In this paper, we propose a real-time emotion recognition approach based on both 2D and 3D facial expression features captured by Kinect sensors. To capture the deformation of the 3D mesh during facial expression, we combine the features of animation units (AUs) and feature point positions (FPPs) tracked by Kinect. A fusion algorithm based on improved emotional profiles (IEPs) and maximum confidence is proposed to recognize emotions with these real-time facial expression features. Experiments on both an emotion dataset and a real-time video show the superior performance of our method.