Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks275 from public data
- Deep Learning for Single Image Super-Resolution: A Brief Review1,154
This survey reviews representative deep learning-based SISR methods and group them into two categories according to their contributions to two essential aspects of S ISR: The exploration of efficient neural network architectures for SISS and the development of effective optimization objectives for deep SISr learning.
- GAN Inversion: A Survey653
This paper provides a survey of GAN inversion with a focus on its representative algorithms and its applications in image restoration and image manipulation, and discusses the trends and challenges for future research.
- Assimilation of remote sensing into crop growth models: Current status and perspectives497
A critique of both the advantages and disadvantages of both EO data and crop growth models is provided, and a solid and robust framework for DA is introduced, where different DA methods are shown to be derived from taking different assumptions in solving for the a posteriori probability density function using Bayes’ rule.
- 479
- Deep metric learning for few-shot image classification: A Review of recent developments237
This paper provides an up-to-date review of deep metric learning methods for few-shot image classification from 2018 to 2022 and categorizes them into three groups according to three stages of metric learning, namely learning feature embeddings, learning class representations, and learning distance measures.
- BSNet: Bi-Similarity Network for Few-shot Fine-grained Image Classification214
A so-called Bi-Similarity Network (BSNet) that consists of a single embedding module and a bi-similarity module of two similarity measures of diverse characteristics that is enabled to learn more discriminative and less similarity-biased features from few shots of fine-grained images, such that the model generalization ability can be significantly improved.
- 188
- 163
- Spatial-Temporal Attention Res-TCN for Skeleton-Based Dynamic Hand Gesture Recognition149
An end-to-end Spatial-Temporal Attention Residual Temporal Convolutional Network (STA-Res-TCN) for skeleton-based dynamic hand gesture recognition, which learns different levels of attention and assigns them to each spatial-temporal feature extracted by the convolution filters at each time step.
- A concise review of recent few-shot meta-learning methods145
This short communication gives a concise review on recent representative methods in few-shot meta-learning, which are categorized into four branches according to their technical characteristics.
- Decorrelation of Neutral Vector Variables: Theory and Applications121
Two fundamental invertible transformations, namely, serial non linear transformation and parallel nonlinear transformation, are proposed to carry out the decorrelation, which shows that a highly negatively correlated neutral vector can be transformed to a set of mutually independent scalar variables with the same degrees of freedom.
- 112
- Adjusting the imbalance ratio by the dimensionality of imbalanced data112
A new imbalance extent metric is proposed, the adjusted IR, by adding a penalty term of the number of discriminative features that is effectively determined by the Pearson correlation test, which demonstrates the effectiveness of theadjusted IR in terms of its better negative correlation with the classification performance.
- $t$-Tests, $F$-Tests and Otsu's Methods for Image Thresholding110
It is naturally demonstrated that the extension of Otsu's binarization method to multi-level thresholding is equivalent to the search for optimal thresholds that provide the largest F -statistic through one-way analysis of variance (ANOVA).
- 106
- 101
- Comment on “On Discriminative vs. Generative Classifiers: A Comparison of Logistic Regression and Naive Bayes”98
Empirical and simulation studies suggest that pairing of either LDA assuming a common diagonal covariance matrix (LDA-Λ) or the naïve Bayes classifier and linear logistic regression may not be perfect, and hence it may be reliable for any claim that was derived from the comparison between LDA- Λ or the naive Bayesclassifier andlinear logistic regressors to be generalised to all generative and discriminative classifiers.
- 96
- 95
- 94
- 90
- 83
- An α-Matte Boundary Defocus Model-Based Cascaded Network for Multi-Focus Image Fusion82
Experiments demonstrate that the proposed MMF-Net outperforms the state-of-the-art methods both qualitatively and quantitatively.
- An integrated method of adaptive enhancement for unsupervised segmentation of MRI brain images82
This paper presents an integrated method of the adaptive enhancement for an unsupervised global-to-local segmentation of brain tissues in three-dimensional MRI (Magnetic Resonance Imaging) images, and yields a robust and accurate segmentation, particularly in noisy images.
- An {\alpha}-Matte Boundary Defocus Model Based Cascaded Network for Multi-focus Image Fusion81
Experiments demonstrate that the proposed MMF-Net outperforms the state-of-the-art methods both qualitatively and quantitatively.
- DREAM: Visual Decoding from REversing HumAn Visual SysteM80
DREAM, an fMRI-to-image method for reconstructing viewed images from brain activities, grounded on fundamental knowledge of the human visual system, outperforms the current state-of-the-art models in terms of the consistency of appearance, structure, and semantics.
- Why Does Rebalancing Class-Unbalanced Data Improve AUC for Linear Discriminant Analysis?80
There is an intrinsic, positive relationship between the rebalancing of class sizes and the improvement of the AUC, and it is shown that the largest improvement of AUC is achieved, asymptotically, when the two classes are fully rebalanced to be of equal sizes.
- 77
- A novel hierarchical framework for human action recognition76
This paper proposes a novel two-level hierarchical framework for 3D skeleton-based action recognition, in order to tackle the challenges of high intra-class variance, movement speed variability and high computational costs of action recognition.
- Do unbalanced data have a negative effect on LDA?75
There is no reliable empirical evidence to support the claim that a (class) unbalanced data set has a negative effect on the performance of linear discriminant analysis (LDA), and re-balancing affects theperformance of LDA for data sets with either equal or unequal covariance matrices, indicating that having unequal covariances matrices is not a key reason for the difference in performance.
- Deep Disturbance-Disentangled Learning for Facial Expression Recognition74
This paper proposes a novel Deep Disturbance-disentangled Learning (DDL) method for FER that is capable of simultaneously and explicitly disentangling multiple disturbing factors by taking advantage of multi-task learning and adversarial transfer learning.
- 73
- Stage-Aware Feature Alignment Network for Real-Time Semantic Segmentation of Street Scenes71
A novel Stage-aware Feature Alignment Network (SFANet) based on the encoder-decoder structure for real-time semantic segmentation of street scenes is presented and an auxiliary training strategy is developed to explicitly alleviate the multi-scale object problem without bringing additional computational costs during the inference phase.
- Ridler and Calvard’s, Kittler and Illingworth’s and Otsu’s methods for image thresholding71
There are close relationships between three popular approaches to image thresholding, namely Ridler and Calvard's iterative-selection (IS) method, Kittler and Illingworth's minimum-error-thresholding (MET) method and Otsu's method.
- TSingNet: Scale-aware and context-rich feature learning for traffic sign detection and recognition in the wild70
A new traffic sign detection network (TSingNet), which learns scale-aware and context-rich features to effectively detect and recognize small and occluded traffic signs in the wild.
- 68
- Real-MFF: A large realistic multi-focus image dataset with ground truth66
A large and realistic multi-focus dataset called Real-MFF is introduced, which contains 710 pairs of source images with corresponding ground truth images, which is generated by light field images, and both the source images and the groundtruth images are realistic.
- 65
- Deep Multi-Task Multi-Label CNN for Effective Facial Attribute Classification64
A novel deep multi-task multi-label CNN, termed DMM-CNN, is proposed for effective FAC, which jointly optimizes two closely-related tasks to improve the performance of FAC by taking advantage of multi- task learning.
- 63
- Customizing 360-Degree Panoramas through Text-to-Image Diffusion Models63
This study proposes an approach that focuses on the customization of 360-degree panoramas, which inherently possess global geometric properties, using a T2I diffusion model, and exhibits exceptional generalization ability in producing scenes unseen in the fine-tuning dataset.
- Advanced Dropout: A Model-free Methodology for Bayesian Dropout Optimization63
The proposed advanced dropout technique applies a model-free and easily implemented distribution with parametric prior, and adaptively adjusts dropout rate, and is extended to uncertainty inference, network pruning, text classification, and regression.
- 63
- Median-based image thresholding60
Two median-based approaches are proposed: one is an extension of Otsu's method and the other is an Extension of Kittler and Illingworth's minimum error thresholding, based on mixtures of Laplace distributions.
- 59
- Image-text dual neural network with decision strategy for small-sample image classification59
This work shows how to efficiently and effectively utilize semantic information of the annotations to improve the performance of small-sample classification by incorporating a decision strategy into the image-text dual neural network to further improve theperformance of the original model on few-shot datasets.
- APANet: Adaptive Prototypes Alignment Network for Few-Shot Semantic Segmentation54
This work presents an adaptive prototype representation by introducing class-specific and class-agnostic prototypes and thus construct complete sample pairs for learning semantic alignment with query features, which effectively enriches feature comparison and helps yield an unbiased segmentation model in the few-shot setting.
- Spectral Nonlocal Restoration of Hyperspectral Images With Low-Rank Property54
A new low-rank spectral nonlocal approach (LRSNL) to the simultaneous removal of a mixture of different types of noises, such as Gaussian noises, salt and pepper impulse noises, and fixed-pattern noises including stripes and dead pixel lines is proposed.
- OSLNet: Deep Small-Sample Classification With an Orthogonal Softmax Layer52
The proposed Orthogonal Softmax Layer (OSL), which makes the weight vectors in the classification layer remain orthogonal during both the training and test processes, has better performance than the methods used for comparison on four small-sample benchmark datasets, as well as its applicability to large-sample datasets.
- Self-reconstruction network for fine-grained few-shot classification49
A self-reconstruction metric module for diversifying query features and a restrained cross-entropy loss for avoiding over-confident predictions are proposed to effectively alleviate overfitting in metric-based few-shot image classification.
- Tilting Methods for Assessing the Influence of Components in a Classifier49
Tilting methods are introduced that can be implemented very rapidly, and it is shown how to use bootstrap methods to assess the accuracy of the variable ranking and variable elimination procedures.
- 46
- Adaptive Deep Disturbance-Disentangled Learning for Facial Expression Recognition45
Extensive experimental results show that the proposed ADDL method consistently outperforms several state-of-the-art FER methods, demonstrating the great potential of disturbance disentanglement for FER.
- 44
- 43
- 42
- 42
- Median-Based Classifiers for High-Dimensional Data41
Using componentwise medians to construct a robust classifier that is relatively insensitive to the difficulties caused by heavy-tailed data and entails straightforward computation is suggested.
- Knowledge-based segmentation and labeling of brain structures from MRI images41
A fuzzy model of regions of interest (ROI) by analogy with the electrostatic potential distribution is introduced, to represent more appropriately the knowledge of distance, shape and relationship of structures.
- 39
- LRID: A new metric of multi-class imbalance degree based on likelihood-ratio test38
A new metric based on the likelihood-ratio test, LRID, is proposed to provide a more reliable measurement of class-imbalance extent for multi-class data.
- Joint sparse model-based discriminative K-SVD for hyperspectral image classification38
A novel dictionary learning method for HSI classification, termed JSM-DKSVD, which can produce a redundant dictionary with rich spectral and spatial information as well as high discriminative power.
- On selecting interacting features from high-dimensional data37
A simple recursive approach to feature-selection methods can lead to substantial improvements in the performance of classifiers, and can provide insight into the way in which features work together in a given population.
- A Survey on Deep Generative 3D-aware Image Synthesis32
This survey aims to introduce new researchers to the task of deep generative 3D-aware image synthesis, provide a useful reference for related works, and stimulate future research directions through the discussion section.
- 32
- 31
- 29
- When Facial Expression Recognition Meets Few-Shot Learning: A Joint and Alternate Learning Framework29
This paper proposes a novel Emotion Guided Similarity Network (EGS-Net), consisting of an emotion branch and a similarity branch, based on a two-stage learning framework, which aims to identify unseen compound expressions with the model trained on easily accessible basic expression datasets.
- 28
- 27
- VPCFormer: A transformer-based multi-view finger vein recognition model and a new benchmark27
A novel Transformer-based model named Vein Pattern Constrained Transformer (VPCFormer) for multi-view finger vein recognition, primarily composed of multiple Vein Pattern Constrained Encoders (VPC-Encoders) and Neighborhood-Perspective Modules (NPMs).
- 27
- 27
- Depth-images-based pose estimation using regression forests and graphical models27
This work proposes a novel 3D Local Shape Context feature extracted from human body silhouette to characterise the local structure of body joints and incorporates a graphical model into regression forests to exploit structural constrains.
- 26
- 25
- Matched Shrunken Cone Detector (MSCD): Bayesian Derivations and Case Studies for Hyperspectral Target Detection25
A new regularized cone-based representation approach to hyperspectral target detection, as well as its two working models by incorporating into the cone representation the matched shrunken cone detector (MSCD).
- 24
- Learning Local Metrics and Influential Regions for Classification24
A new intuitive distance is defined with local metrics and influential regions, and subsequently a novel local metric learning algorithm called LMLIR is proposed for distance-based classification.
- 23
- RFRN: A recurrent feature refinement network for accurate and efficient scene text detection23
RFRN, as a recurrent segmentation framework, contains a recurrent path augmentation that refines the previous feature maps as inner states, which not only helps improve the segmentation quality, but also fully facilitates the reuse of parameters and low computational cost.
- 23
- MvSSIM: A quality assessment index for hyperspectral images23
A new and simple multivariate SSIM (MvSSIM) index for HSI is proposed, by treating the pixel spectrum as a multivariate random vector and maintains SSIM’s ability to assess the spatial structural similarity via correlation between two images of the same band; and adds an ability to assessment the spectral structural similarity through covariance among different bands.
- 22
- Cali-sketch: Stroke calibration and completion for high-quality face image generation from human-like sketches22
Cali-Sketch is proposed, a human-like-sketch to photo-realistic-image generation method that decouple a difficult cross-domain translation problem into two easier steps and is compared with state-of-the-art methods.
- 22
- 21
- The neglected background cues can facilitate finger vein recognition21
This paper proposes a new descriptor named Intensity Orientation Vector (IOV), scaleable to reflect characteristics of finger tissues, and proposes a new learning scheme named Semantic Similarity Preserved Discrete Binary Feature Learning (SSP-DBFL) for vein recognition.
- Channel-Spatial Support-Query Cross-Attention for Fine-Grained Few-Shot Image Classification20
A new Channel-Spatial Cross-Attention Module (CSCAM) is proposed, which can effectively drive a model to extract discriminative fine-grained feature representations with only few shots, conveniently enabling effective improvement of state-of-the-art methods for few-shot fine-grained image classification.
- 20
- Data-augmented matched subspace detector for hyperspectral subpixel target detection20
This paper first synthesise target-background mixed spectra based on classical hyperspectral mixing models and then learn a target- Background mixed subspace via principal component analysis, which is more representative as spectral variability of target spectra is explained to the largest extent and it leads to an improvement in computational speed and numerical stability.
- 20
- 19
- ReMarNet: Conjoint Relation and Margin Learning for Small-Sample Image Classification19
This paper proposes to enhance the discriminative power of features from a new perspective by introducing a novel neural network termed Relation-and-Margin learning Network (ReMarNet), which assembles two networks of different backbones so as to learn the features that can perform excellently in both of the aforementioned two classification mechanisms.
- Embedding metric learning into set-based face recognition for video surveillance19
A novel method which embeds distance metric learning into set-based image matching, which can adapt the variation in viewpoints and illuminations across cameras in order to improve face recognition in video surveillance.
- 18
- 18
- Defocus map estimation from a single image using improved likelihood feature and edge-based basis18
This paper proposes a DME method combining edge-based and region-based methods together to keep their respective advantages while eliminating the shortcomings, and demonstrates that the proposed method outperforms state-of-the-art DME methods.
- Controllable Continuous Gaze Redirection18
This work designs a novel framework for controllable gaze redirection that achieves both precise redirection and continuous interpolation and introduces a high-quality gaze image dataset with a large range of directions, which also benefits researchers in related areas.
- Enhanced Grassmann discriminant analysis with randomized time warping for motion recognition18
The validity of the proposed framework, RTW+eGDA, is demonstrated through experiments on motion recognition using the publicly available Cambridge gesture, KTH action, and UCF sports datasets.
- Rayleigh-distribution based minimum error thresholding for SAR images18
This paper presents a minimum error thresholding (MET) algorithm under the hypothesis that the gray level histogram of SAR image fits to a mixture model of shifted Rayleigh distribution and compared with traditional Otsu algorithm and other MET algorithms based on various models of histogram.
- 17
- 17
- Unsupervised learning of pixel clustering in Mueller matrix images for mapping microstructural features in pathological tissues17
It is confirmed that such unsupervised learning technique can decompose the histological structures into a stable basis of characteristic microstructural clusters, some of which correspond to distinctive pathological features for clinical diagnosis.
- Polarization-based probabilistic discriminative model for quantitative characterization of cancer cells17
A polarization-based probabilistic discriminative model is proposed for deriving a set of new sigmoid-transformed polarimetry feature parameters, which not only enables accurate and quantitative characterization of cancer cells at pixel level, but also accomplish the task with a simple and stable model.
- D³Net: Dual-Branch Disturbance Disentangling Network for Facial Expression Recognition17
This paper proposes a novel Dual-branch Disturbance Disentangling Network (D3Net), mainly consisting of an expression branch and a disturbance branch, to perform effective FER and introduces an Indian buffet process (IBP) prior to model the distribution of potential disturbing factors in LFS, thereby enhancing the disentanglement of disturbing factors.
- Unsupervised multi-domain multimodal image-to-image translation with explicit domain-constrained disentanglement17
This paper proposes a unified framework for learning to generate diverse outputs using unpaired training data and allow for simultaneous multi-domain translation via a single model and investigates how to extract domain-level signal as explicit supervision so as to achieve better image-to-image translation.
- LCSCNet: Linear Compressing-Based Skip-Connecting Network for Image Super-Resolution17
A linear compressing layer is designed in LCSCNet for skip connection, which connects former feature maps and distinguishes them from newly-explored feature maps, and an adaptive element-wise fusion strategy with multi-supervised training is proposed.
- Randomized time warping for motion recognition17
A randomized extension of the DTW concept, termed randomized time warping (RTW), for motion recognition, which simultaneously computes multiple degrees of similarities between a number of warped patterns' pair candidates and generalizes the Hankel matrix commonly used in modeling of system dynamics.
- 16
- Inter-class angular margin loss for face recognition16
The proposed Inter-class angular margin (IAM) loss adaptively penalizes smaller inter-class angles more heavily and successfully makes the angular margin between classes larger, which can significantly enhance the discrimination of facial features.
- 16
- A hybrid finger identification pattern using Polarized depth-Weighted Binary Direction Coding16
The proposed PWBDC method performs well on both the newly established database of hybrid images and a popular public database of traditional finger vein images, superior to many established and state-of-the-art methods.
- Combining multiple biometric traits with an order-preserving score fusion algorithm16
A novel order-preserving probabilistic score fusion algorithm, Order-Preserving Tree (OPT), is proposed, by casting the score fusion problem into an optimisation problem with the natural order- Preserving constraint.
- 16
- 15
- 15
- Cali-Sketch: Stroke Calibration and Completion for High-Quality Face Image Generation from Poorly-Drawn Sketches15
Cali-Sketch is proposed, a poorly-drawn-sketch to photo-realistic-image generation method that decouple a difficult cross-domain translation problem into two easier steps.
- MiLDA: A graph embedding approach to multi-view face recognition15
This paper proposes a graph embedding framework that can project the multi-view data into a common subspace of higher discriminability between classes, and demonstrates that MiLDA has an excellent performance and outperforms many popular approaches.
- On the generative–discriminative tradeoff approach: Interpretation, asymptotic efficiency and classification performance15
The interpretation of generative, discriminative and hybrid approaches to classification is discussed, in particular for the generative-discriminative tradeoff (GDT), a hybrid approach, and the asymptotic efficiency of the GDT, relative to that of its generative or discrim inative counterpart, is presented theoretically and numerically.
- Constrained mutual convex cone method for image set based recognition14
A convex cone based framework is established, which mathematically defines multiple angles between two convex cones, and then defines the geometric similarity between the cones using the angles, and a discriminant space is introduced that maximizes the between-class variance (gaps) and minimizes the within- class variance (Gaps) on the discriminantspace.
- 13
- 13
- How Video Super-Resolution and Frame Interpolation Mutually Benefit13
This work designs a spatial-temporal super-resolution network based on exploring the interaction between VSR and VFI and achieves state-of-the-art performance on various datasets.
- DS-UI: Dual-Supervised Mixture of Gaussian Mixture Models for Uncertainty Inference in Image Recognition13
Experimental results show the DS-UI outperforms the state-of-the-art UI methods in misclassification detection and finds statistically significant improvements in open-set out- of-domain/-distribution detection.
- Cone-based joint sparse modelling for hyperspectral image classification13
Experiments and investigations show that the proposed C-JSM can produce a more stable, sparse representation and a superior classification than other methods which only ensure the sparsity, non-negativity or spatial coherence.
- Matched shrunken subspace detectors for hyperspectral target detection13
Experiments demonstrate that the proposed MSSD methods can outperform the original MSD for hyperspectral target detection and provide both the frequentist and Bayesian derivations.
- 12
- 12
- Bi-Stream Pose-Guided Region Ensemble Network for Fingertip Localization From Stereo Images12
A novel scheme termed bi-stream pose-guided region ensemble network (Bi-Pose-REN), which extracts more representative feature regions around joints in the feature maps under the guidance of the previously estimated pose, to better present the performance of fingertip localization on THU-Bi-Hand.
- 12
- 11
- Query-centric distance modulator for few-shot classification11
A QCDM is proposed to generate query-related weights for each query sample to address the issue of similarity between samples.
- Discriminatively guided filtering (DGF) for hyperspectral image classification11
The original HSI is projected into the low-dimensional space induced by LDA to serve as guidance images for filtering the intermediate classification results induced by SVM, and experiment results show the superior performance of the proposed DGF compared with that of the principal component analysis (PCA)-based GF.
- On the orthogonal distance to class subspaces for high-dimensional data classification11
It is shown that an established and widely-used relationship, between the residual standard deviation and the sum of squares of the residual PC scores, is not precise, and thus can lead to incorrect results, for the inference of high-dimensional data which nowadays are common in practice.
- 11
- Joint discriminative–generative modelling based on statistical tests for classification11
Numerical experiments demonstrate that the performance of this new approach to classification is similar to or better than that of its discriminative and generative counterparts, in particular when the size of the training-set is comparable to the dimension of the data.
- Consistency-driven feature scoring and regularization network for visible–infrared person re-identification10
A novel consistency-driven feature scoring and regularization network (CFSR-Net) is proposed, which consists of a backbone network, a local feature learning block, a feature scoring block, and a global-local feature fusion block, for VI-ReID.
- 10
- ASSP: An adaptive sample statistics-based pooling for full-reference image quality assessment10
This paper develops an innovative pooling strategy based on sample statistics to adaptively make the IQA more consistent with human visual assessment by adaptively integrating robust sample statistics and standard sample statistics via excess kurtosis.
- 10
- 10
- Hyperspectral image denoising based on low-rank coefficients and orthonormal dictionary10
A novel HSI denoising model based on Low-Rank Coefficients and Orthonormal Dictionary (LRCOD) is proposed, which yields the best MPSNR in 30 of 35 cases and runs faster than other methods except one non-iterative algorithm.
- 10
- The matched subspace detector with interaction effects10
The MSDinter introduces interaction effects terms into the popular matched subspace detector (MSD), from regression analysis in multivariate statistics and the bilinear mixing model in hyperspectral unmixing, such that superior performance of target detection can be achieved.
- Building a discriminatively ordered subspace on the generating matrix to classify high-dimensional spectral data10
A discriminatively ordered subspace (DOS) is proposed: different from GDS, this DOS selects the eigenvectors with high discriminative ability between classes rather than between class subspaces, demonstrating that applying DOS before SIMCA outperforms its counterparts.
- 10
- Uncertainty-Aware Label Refinement on Hypergraphs for Personalized Federated Facial Expression Recognition9
Experimental results on two challenging real-world facial expression databases show that the proposed method consistently outperforms several state-of-the-art methods, indicating the superiority of hypergraph modeling for uncertainty estimation and label refinement on the personalized federated FER task.
- 9
- Deep InterBoost networks for small-sample image classification9
Experimental results on four small-sample datasets, UIUC-Sports, LabelMe, 15Scenes and Caltech101, demonstrate that the proposed ensemble method outperforms existing ones and results from the Wilcoxon signed-rank tests show that the method is statistically significantly better than the methods compared.
- 8
- 8
- Augmentation Matters: A Mix-Paste Method for X-Ray Prohibited Item Detection Under Noisy Annotations8
This paper addresses the challenging problem of training a robust prohibited item detector under noisy annotations from a novel perspective of data augmentation, and proposes an effective label-aware mixed patch paste augmentation method (Mix-Paste), which demonstrates the great potential of data augmentation to handle noise annotations.
- SPL-Net: Spatial-Semantic Patch Learning Network for Facial Attribute Recognition with Limited Labeled Data8
Experimental results on challenging facial attribute datasets show the superiority of SPL-Net over several state-of-the-art methods in the case of limited labeled data.
- 8
- 8
- 8
- On the Comparisons of Decorrelation Approaches for Non-Gaussian Neutral Vector Variables8
This article extensively compare PNT with PCA and ICA through both theoretical analysis and experimental evaluations, and demonstrates the superiority of PNT for decorrelating the neutral vector variables.
- Learning distance to subspace for the nearest subspace methods in high-dimensional data classification8
This paper proposes to improve the classification performance of NSM through learning tailored distance metrics from samples to class subspaces through ‘learned distance to subspace’ (LD2S) and shows superior classification performance on three real-world high-dimensional spectral datasets.
- No-reference image quality assessment for photographic images based on robust statistics8
Experimental results show that RSN remarkably outperforms state-of-the-art NR-IQA methods on the CID2013 database of photographic images, as well as on the popular LIVE and TID2013 databases.
- What is the probability of replicating a statistically significant association in genome-wide association studies?8
The Bayesian replication probability is referred as the replication rate (RR) and an estimation method for RR is described, which makes use of the summary statistics from the primary study, and an R-package is described to estimate and apply RR in GWASs.
- Making XML document markup international8
This paper presents an XSLT‐based method that can completely localize the markup of XML documents into different natural languages and describes how the proposed technique can be applied to translation problems in programming or documentation so that a program or a document can be converted to and from an XML format.
- A brief review of deep learning methods in mortality forecasting7
This paper provides the first concise view of deep learning in mortality forecasting by covering six deep network architectures, namely Recurrent Neural Networks, Convolutional Neural Networks, Convolutional Neural Networks, Transformers, Autoencoders, Locally Connected Networks, and Multi-Task Feed-Forward Networks.
- Automatic 3D joint erosion detection for the diagnosis and monitoring of rheumatoid arthritis using hand HR-pQCT images7
This paper defines erosion as the significant concave regions on the cortical layer, and develops a model-based 3D automatic erosion detection method that has shown satisfactory and consistent performance compared with the annotations provided by medical experts.
- 7
- Latent variable pictorial structure for human pose estimation on depth images7
A latent variable-based prior model for human pose estimation that adaptively adopts prior pose models based on the estimated value of the latent variable and enables the learning of a more accurate part classifier is developed.
- 7
- Interpretation of hybrid generative/discriminative algorithms7
It is suggested that the hybrid algorithm is by nature a generative model with its parameters learnt through both generative and discriminative approaches, in the sense that it assumes a scaled data-generation process and uses scaled class-posterior probabilities to perform discrimination.
- EgoTactile: Learning Grasp Pressure for Everyday Objects from Egocentric Video6
This work introduces EgoTactile, a benchmark pairing egocentric video with full-hand pressure supervision for diverse everyday objects, incorporating a bare-hand transfer subset to enable generalization to natural scenarios and achieves superior performance on the benchmark and robust transferability to in-the-wild scenarios.
- 6
- 6
- 6
- 6
- 6
- 6
- 6
- 6
- Fisher׳s linear discriminant embedded metric learning6
A novel distance metric learning model is proposed, which embeds [email protected]?s linear discriminant into the classical maximum margin criterion, which benefits from exploiting the distributions of sample pairs and thus becomes more reliable and effective.
- 6
- Threshold selection using cross-entropy and fuzzy divergence6
This paper presents four new algorithms for optimal threshold selection based on different criteria integrating cross entropy and fuzzy divergence, and constructs a new fuzzy membership function to take into account the gray level probability distribution of object pixels and background pixels about their mean values for the last two algorithms.
- Interactive triplet attention for few-shot fine-grained image classification5
This paper shows that a simple adjustment to the existing triplet attention module (TAM) can be highly effective for few-shot fine-grained image classification, and introduces an additional interaction between the outputs of these three branches, capable of modeling the dependency across all three dimensions.
- 5
- Toward Certified Robustness of Distance Metric Learning5
This article shows that the enlarged margin is beneficial to the generalization ability by using the theoretical technique of algorithmic robustness and demonstrates the superiority of the proposed method over existing state-of-the-art methods in both discrimination accuracy and robustness against possible noise.
- 5
- 5
- 5
- Generalization Bound of Gradient Descent for Non-Convex Metric Learning5
This paper theoretically address this question and proves the agnostic Probably Approximately Correct (PAC) learnabil-8 ity for metric learning algorithms with non-convex objective functions optimized via gradient descent (GD); in particular, the theoretical guarantee takes training time into account.
- 5
- MSDH: Matched subspace detector with heterogeneous noise5
Experiments on two benchmark real hyperspectral datasets demonstrate the superiority of MSDH over MSD for subpixel target detection.
- BALSON: BAYESIAN LEAST SQUARES OPTIMIZATION WITH NONNEGATIVE L1-NORM CONSTRAINT5
With the estimated posterior distributions, the original parameters can be effectively reconstructed in polynomial fitting problems, and the BALSON framework is found to perform better than conventional methods.
- 360PanT: Training-Free Text-Driven 360-Degree Panorama-to-Panorama Translation4
The proposed 360PanT is a training-free approach to text-based 360-degree panorama-to-panorama translation with boundary continuity that achieves seamless translations through two key components: boundary continuity encoding and seamless tiling translation with spatial control.
- FETNet: Feature Exchange Transformer Network for RGB-D Object Detection4
A Feature Exchange Transformer Network (FETNet), which consists of two well-designed components: the Feature Exchange Module (FEM), and the Multi-modal Vision Transformer (MViT) to effectively fuse multi-modal features and exploit the attention between the RGB and depth features.
- 4
- SEA: A Combined Model for Heat Demand Prediction4
STL-ENN-ARIMA (SEA), a combined model, was proposed based on the combination of the Elman neural network and the autoregressive integrated moving average model, which are commonly applied to heat demand prediction.
- 4
- Incorporating prior probabilities into high-dimensional classifiers4
A general approach to reducing error rate is suggested by using a method derived from Breiman's bagging idea, which is simple and explicit to apply, and does not involve choice of any tuning parameters.
- A Survey on Text-Driven 360° Panorama Generation3
This survey presents a comprehensive review of text-driven 360-degree panorama generation, offering an in-depth analysis of state-of-the-art algorithms and critically examine current limitations and propose promising directions for future research.
- 3
- Modelling Latent Dynamics of StyleGAN using Neural ODEs3
The entire sequence is seen as discrete-time observations of a continuous trajectory of the initial latent code, by considering each latent code as a moving particle and the latent space as a high-dimensional dynamic system.
- 3
- Small-Vote Sample Selection for Label-Noise Learning3
This paper proposes a novel yet simple sample selection method, which mainly consists of a Hierarchical Voting Scheme (HVS) and an Adaptive Clean data rate Estimation Strategy (ACES), to accurately identify clean samples and noisy-labeled samples for robust learning.
- Guest Editorial Special Issue on Recent Advances in Theory, Methodology, and Applications of Imbalanced Learning3
The aim of this Special Issue is to collect recent research works that focus on the theory, methodology, and applications of imbalanced learning.
- 3
- Simple tiered classifiers3
Empirical studies show that three two-tier classifiers, which are respective extensions of linear discriminant analysis, linear logistic regression and support vector machines, can reduce noticeably the relatively high misclassification error of their original single-tier counterparts, without significantly increasing computational labour.
- Fine-grained mortality forecasting with deep learning2
MortFCNet is introduced, a deep-learning method that predicts weekly death rates using region-specific weather inputs using region-specific weather inputs that consistently outperforms both a standard Serfling-type baseline and XGBoost in terms of predictive accuracy.
- 2
- BDC-Occ: Binarized Deep Convolution Unit for Binarized Occupancy Network2
A novel binarized deep convolution unit is proposed that significantly enhances performance, even when the number of binarized convolutional layers increases to meet the requirements of 3D occupancy networks, and minimizes binarization errors and improves perceptual capability.
- DARL: Mitigating Gradient Conflicts in Long-Tailed Out-of-Distribution Learning2
Dynamic Ambiguity-aware Recalibration for Logits (DARL) is proposed, an ambiguity-guided long-tailed OOD learning approach that introduces an ambiguity-aware logit adjustment that can dynamically calibrate the class margins using energy-based ambiguity metrics, effectively reducing early-stage bias while avoiding late-stage overfitting.
- 2
- Reschedule Diffusion-based Bokeh Rendering2
A prior-aware sampling approach, which can adaptively control the noise scale through learned priors, and a prior-aware noise scheduling strategy, which can greatly reduce the number of inference steps without sacrificing performance are proposed.
- Real-MFF Dataset: A Large Realistic Multi-focus Image Dataset with Ground Truth.2
A large and realistic multi-focus dataset containing 800 pairs of source images with the corresponding ground truth images, generated using a light field camera to serve as a well-founded benchmark for multi- focus image fusion tasks.
- Special Issue on Advances in Statistical Methods-based Visual Quality Assessment2
This special issue collected eight papers reporting the state-of-the-art researches in the theory, algorithm, modeling, system and database of statistical methods-based VQA and demonstrating the latest efforts of relevant researchers.
- 2
- Deformable Part-Based Model Transfer for Object Detection2
A new strategy to transfer deformable part models (DPMs) for object detection is proposed, using offline-trained auxiliary DPMs of similar categories as source models to improve the performance of the target object detector.
- Aspects of generative and discriminative classifiers2
This thesis proposed a joint generative-discriminative modelling approach to classification and claimed that there exist two distinct regimes of performance between the generative and discriminative classifiers with regard to the training-set size; however, empirical and simulation studies suggest that it is not so reliable to claim such an existence of the two separate regimes.
- Localizing XML Documents through XSLT.2
This paper presents a pure XSLT stylesheet to completely translate and localize XML documents into different natural languages and describes how this technique can be applied to translation problems in programming or documentation languages when the program or the document can be converted to and from XML documents.
- DYNAMIC IMAGE SEGMENTATION USING 2-D GENETIC ALGORITHMS2
Two new algorithms for dynamic image segmentation based on 2 D genetic algorithms based on fuzzy set theory and energy function of Hopfield neural network have superior performance for noisy images.
- Federated learning with noisy labels: A comprehensive and concise review of current methodologies and future directions1
This work offers the first comprehensive and concise review dedicated to FLNL and proposes a taxonomy to categorise current FLNL studies into four types that address the four challenges correspondingly: sample-wise methods, client-wise methods, model-wise methods, and benchmark-wise studies.
- Class-Aware Cross-Attention Helps Prompt Learning With Limited Samples1
The proposed class-aware cross-attention prompt learning (CAPrompt) is evaluated on 11 widely used datasets and the results showcase its effectiveness in achieving the state-of-the-art performance of image classification with limited samples.
- 1
- VISTA: Unsupervised 2D Temporal Dependency Representations for Time Series Anomaly Detection1
VISTA is introduced, a training-free, unsupervised TSAD algorithm designed to overcome challenges of intricate temporal relationships in time series data, and achieves state-of-the-art performance on five multivariate TSAD benchmarks.
- 1
- 1
- 1
- 1
- DCES-PA: Deformation-controllable elastic shape model for 3D bone proliferation analysis using hand HR-pQCT images1
An unsupervised method, named Deformation-Controllable Elastic Shape model, for 3D bone Proliferation Analysis (DCES-PA), which not only shows an improved accuracy than other state-of-the-art shape-based methods applied to proliferation analysis but also produces highly sensitive proliferation annotations to assist physicians in proliferation analysis.
- GKF-PUAL: A group kernel-free approach to positive-unlabeled learning with variable selection1
The experimental results reveal the superiority of GKF-PUAL in both PU classification and variable selection, improving the baseline PUAL by more than 10% in F1-score across four benchmark datasets and removing over 70% of irrelevant variables on six benchmark datasets.
- PUAL: A Classifier on Trifurcate Positive-Unlabeled Data1
A PU classifier with asymmetric loss (PUAL) is proposed, by introducing a structure of asymmetric loss on positive instances into the objective function of the global and local learning classifier, and a kernel-based algorithm is developed to enable PUAL to obtain non-linear decision boundary.
- 1
- 1
- Lane Line Detection based on Parallel Spatial Separation Convolution1
This work proposes Parallel Spatial Separation Convolution with Message-Passing (PSS- Conv), a new convolution operation built on a new parallel spatial convolution decomposition and a channel-weighted feature merging strategy, to aggregate the features obtained from decomposed convolution.
- 1
- Guest Editorial: Special issue on deep learning with small samples1
This special issue collects the latest developments in various aspects of deep learning with small samples and provides a concise review on recent representative methods in few-shot meta-learning, which are categorised into four branches according to their technical characteristics.
- 1
- 1
- 1
- Detection of user-registered dog faces1
A novel user- registered object detection framework that can generate an adaptive detector, from only a limited number of user-registered target samples and a couple of off-line trained auxiliary models is proposed.
- Short note on two output-dependent hidden Markov models1
The purpose of this note is to study the assumption of ''mutual information independence'', which is used by Zhou for deriving an output-dependent hidden Markov model, the so-called discriminative HMM (D-HMM), in the context of determining a stochastic optimal sequence of hidden states.
- 1
- Brain tissue classification based on a mixel model and Markov random field models1
A fully-automatic 3D classification of brain tissues for Magnetic Resonance (MR) images using Markov Random Field (MRF) models and the multifractal dimension, describing the topology of the brain, is added to the MRF’s to improve the discrimination of the mixclasses.
- CoDA: Co-adaptive Dual-path Alignment for Vision-Language Models–
In CoDA, a new adaptation framework that explicitly disentangles and coordinates cross-modal semantic alignment and intra-modal structural consistency is proposed, and it is shown that CoDA outperforms state-of-the-art parameter-efficient methods, particularly under few-shot learning and distribution-shift scenarios.
- –
- Simple Domain Generalization for Strong Pixel-Level Image Tampering Detection in Modern VLMs–
This work introduces a balanced minibatch sampling scheme that strategically samples tampered and real images in each minibatch, preventing biased optimization toward either manipulated artifacts or clean-image priors and avoiding training collapse, and adopts a simple late-injection strategy.
- –
- Probing CLIP's Comprehension of 360-Degree Textual and Visual Semantics–
A LoRA-based fine-tuning framework that explicitly instills invariance to circular shifts is proposed that exhibits improved comprehension of 360-degree visual semantics, though with a slight degradation in original semantic evaluation performance, highlighting a fundamental trade-off in adapting CLIP to 360-degree panoramic images.
- –
- Change Only What You Need: Dual Dynamic Hypergraph-Driven Fusion for GAN Inversion–
This work proposes a novel dual dynamic hypergraph-driven fusion (DHFusion) method, which consists of a dual dynamic hypergraph-CAM network (DH-Net) and an editing-driven fusion network (EF-Net), and shows the potential of dual dynamic hypergraphs for GAN inversion.
- Pose Magic++: Integrating Mamba and HyperGCN for Efficient and Temporally Consistent 3D Human Pose Estimation–
This work proposes a novel attention-free integrated spatiotemporal architecture named Integrated Mamba-HyperGCN (Pose Magic++), which significantly improves the estimation accuracy of peripheral joints without incurring extra computational cost.
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- Automatic Detection of 3D Joint Erosion in Hand HR-pQCT: A Level Set-Based Approach–
This paper proposes a 3D fully automatic joint erosion algorithms for hand HR-pQCT, and its performance can reach consensus with human specialist.
- –
- –
- –
- Generalisations of stochastic supervision models–
This paper provides four generalisations of stochastic supervision models, extending them to asymmetric assessments, multiple classes, feature-dependent assessments and multi-modal classes, respectively, and derives four new EM algorithms.
- Towards Certified Robustness of Metric Learning–
This work advocates penalizing small distance between training instances and their nearest adversarial examples, and shows that the resulting new approach to metric learning enjoys a larger certified neighborhood with theoretical performance guarantee.
- –
- –
- –
- A new approach to robust fundamental matrix estimation using an analytic objective function and adjusted gradient projection–
Experimental results show that the proposed approach performs better than traditional robust methods RANSAC, MSAC, LMEDS and MLESAC, in particular when correspondences were seriously corrupted.
- Domain-Aware No-Reference Image Quality Assessment.–
The domain-aware no-reference image quality assessment (DA-NR-IQA) is proposed, which for the first time exploits and disentangles the distinct representation of different degradations to access image quality.
- –
- –
- –
- –
- –
- Brain Structure Classification In MRI Images–
An automatic algorithm is presented to segment and classify neuroanatomic brain structures in MRI brain images, using the structural knowledge derived from Talairach stereotaxic atlas, as well as the statistical and multifractal information from MRI images.
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-15. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.