Spurious correlations cause deep learning models to rely on predictive shortcuts that hold in the training data but break under distribution shifts, leading to large performance drops for minority groups. Existing strategies often rely on costly group annotations or employ unstable adversarial training. In this paper, we propose Prototype-guided debiasing using Robust Invariant Feature Transformations (PRIFT), a novel framework that mitigates spurious correlations by manipulating latent space geometry. Specifically, we introduce a prototype-guided modeling approach that leverages natural language prompts to represent confounders, transforming abstract biases into interpretable geometric anchors without auxiliary classifiers. Based on these anchors, we introduce a centered projection operator that adaptively purifies representations by removing confounding deviations specific to instances while preserving essential semantic structure. Furthermore, PRIFT can handle confounding factor information at different levels, ranging from true labels to unsupervised latent inference. Experiments on four text classification benchmarks demonstrate the superiority of our method; notably, PRIFT outperforms state-of-the-art baselines and improves worst-group accuracy by over 20% on the CivilComments dataset compared to standard empirical risk minimization.

Aspect-based sentiment analysis (ABSA) aims to extract fine-grained opinions from the text by discerning sentiments toward specific aspects. Although large language models (LLMs) perform well in-context learning (ICL), current ICL methodologies typically retrieve semantically similar but structurally redundant examples, failing to capture syntactic and aspect-level cues critical for ABSA. To overcome these limitations, we report Multi-perspective Sequential retrieval with Predictive Feedback (MSPF), a few-shot learning framework that enhances ICL through MSPF, which integrates three complementary perspectives: overall semantic, syntactic relevance, and aspect sentiment alignment. Evaluated on four benchmark datasets (Laptop14, Restaurant14, Books, and Clothing), MSPF achieved F1 scores of 67.03 % (Laptop14), 73.51 % (Restaurant14), 76.07 % (Books), and 81.96 % (Clothing), outperforming standard ICL by +7.06 %, +5.60 %, +25.61 %, and +18.38 %, respectively. These results validated the efficacy of MSPF in improving LLM reasoning for fine-grained sentiment tasks with limited annotations.

High-dimensional and weakly supervised (HiDWS) data present significant challenges for traditional machine learning and pattern recognition. Although semi-supervised feature selection has shown effectiveness in improving the quality of HiDWS data, existing methods remain sensitive and lack robustness due to the unreliability of unlabeled data learning and the uncertainty in modeling processes. Hence, this study focuses on a multi-granularity zentropy modeling (Ze-MGM) framework with model-agnostic for highly-accuracy and robust semi-supervised feature selection. Unlike existing methods, Ze-MGM does not rely on specific settings such as rough or fuzzy set assumptions and can effectively capture the granularity of information under high-dimensional and weakly supervised data scenarios. Specifically, we first introduce a strategic soft label ($S2-$S2-Label) learning method that integrates object proximity and classification certainty to reduce uncertainty between features and labels. This method also enables the selection of compatible instances, thereby mitigating the negative impact of incompatible objects on label learning. Subsequently, a multi-granularity knowledge space and zentropy uncertainty measure are constructed by analyzing the hierarchical relationships among labels, decisions, and specific classes, which enables accurate multi-granularity knowledge representation and multi-granularity uncertainty characterization in HiDWS data modeling processing. Finally, two multi-granularity significance measures based on multi-granularity uncertainty are defined for feature evaluation and selection via a semi-supervised paradigm. Extensive experiments on multiple benchmark datasets demonstrate that the proposed Ze-MGM method achieves superior generalization performance and robustness compared to state-of-the-art methods.

Camouflaged Object Detection (COD) is a formidable computer vision challenge due to the striking resemblance between camouflaged objects and their surroundings. Despite progress in existing methods, they still face significant limitations, particularly in addressing the issues of fuzzy boundaries and the inadequate fusion of local and global features. To address these challenges, we present a multi-scale COD network named Multi-Scale Local-Global Fusion (MSLGF). MSLGF incorporates a Multi-Scale Fusion Module (MSFM), which skillfully integrates feature maps at multiple scales to produce high-fidelity edge features. Additionally, to refine the detection process, a Local-Global Feature Fusion Module (LGFFM) combines the local edge details with global semantic information of camouflaged targets, significantly enhancing the accuracy of COD. Experimental results show that MSLGF achieves remarkable performance across 3 benchmark datasets, i.e., Camouflaged Object Dataset (CAMO), Camouflaged Object Dataset with 10,000 Images (COD10K), and NC4K. Specifically, MSLGF attains a structure-measure from 0.879 to 0.894 and a weighted F-measure between 0.817 and 0.856. The source code is publicly available at https://github.com/tc-fro/MSLGF.

Multigranularity knowledge modeling is an influential study for information processing and knowledge discovery in artificial intelligence (AI). A central research focus is the multigranularity representation and learning of knowledge structures. Among them, fuzzy rough sets (FRSs) have emerged as a representative method for characterizing uncertain knowledge. However, the existing FRS studies still exhibit two limitations: low robustness in knowledge acquisition and incomplete characterization of uncertainty. Hence, this article proposes a zentropy-enhanced multigranularity knowledge modeling framework for robust feature selection (ZeMG-FS). Specifically, we design a fast and adaptive multigranularity information granulation mechanism based on generalized granular-ball generation to effectively capture data distributions embedded in complex data. Then, the fuzzy rough approximation method is incorporated into the representation of multigranularity knowledge. Furthermore, we analyze the fundamental relationships and structures of the multigranularity knowledge model to introduce a novel multilevel zentropy. Unlike existing entropy measures, the primary consideration of the proposed zentropy is to match and enhance the performance of the proposed model. Finally, we design two feature evaluation criteria grounded in the model and apply them to feature selection. Extensive experiments demonstrate that our proposed methods achieve superior robustness and effectiveness compared with state-of-the-art approaches.

Salient object detection (SOD) aims to distinguish salient regions from non-salient ones in an image. In real-world scenarios, factors such as depth variation and surface reflection can interfere with the model's judgment, while illumination uncertainty further intensifies this interference. As a result, the uncertainty in salient boundary detection increases, leading to false or missed detections. To core with uncertainty inherent to the problem, we introduce the concept of shadowed set, which is an effective method to process the uncertainty problem. In this paper, we have designed an illumination-aware feature integration network by conducting dual-input feature integration under the implicit assistance of illumination maps. Firstly, we devised a determination of pixel-level salient area module, which extract illumination maps based on Retinex theory and obtain the main area of salient object based on shadowed set as the implicit feature of illumination. Next, we constructed a dual-modal compression module to solve the problem of feature alignment, which can use the dual-stream structure to process RGB and auxiliary inputs. Finally, multi-stage contextual complementary module can effectively recover fine object edges, and we use the outputs from the last three stages to supervise the training of the entire model. The state originality came from our previous work on illumination maps and shadowed sets, and we creatively combined them with the SOD to process uncertainty of salient area. The experiments demonstrate that our method exhibits excellent performance on multiple RGB-based datasets, at the same time, it also demonstrates unique performance on underwater and challenging scenes.

Feature selection is a crucial data preprocessing technique aimed at enhancing robustness against boundary uncertainty and complex geometric distributions and improving model interpretability by removing irrelevant or redundant features. Most filter-based feature selection methods employ a single evaluation metric and often fail to explore the underlying geometric distribution structure of the data, which holds significant guiding information for feature selection. In view of this, this paper proposes a Supervised Feature Selection framework guided by granular-ball computing (SFS-GBC). The framework aims to identify superior feature subset candidates for downstream classification tasks by leveraging geometric insights derived from granular balls. Meanwhile, the proposed algorithm introduces a unique two-stage evaluation process designed to synergize coarse-grained efficiency with fine-grained geometric precision, adapting to data of varying dimensionality. Spectifically First, Granularity Consistency (GC) is employed to rapidly rank all features, effectively pre-screening for candidates highly relevant to the decision classes. More innovatively, the subsequent stage departs from traditional statistical metrics by proposing a novel evaluation measure based on the geometric distribution of granular balls, combining Intra-class Compactness (ICC) and Inter-class Discrimination (ICD) to accurately assess feature subsets within a forward search procedure. Furthermore, by adjusting a granulation parameter, the algorithm can explore the data structure from a multi-granularity perspective, potentially yielding superior feature subsets. Experimental results on public benchmark datasets demonstrate the superiority and effectiveness of the proposed approach.

Occluded person re-identification aims to address the identification challenges posed by pedestrians obscured by other individuals or objects. Existing methods often rely on incorporating pose or semantic information to improve model performance under occlusion. However, such information often depends on external models with inevitably cross-domain gaps, whose stability is limited in complex occlusion environments and prone to false results. In this paper, we propose a Transformer-based uncertainty-driven Gaussian model, termed as UD-Gaussian. Firstly, to enrich the detailed features of pedestrian images, a high-frequency enhancement module is introduced. The high-frequency components of the pedestrian image are extracted by Discrete Haar Wavelet Transform, and Top-K high-frequency patches are extracted to construct a graph Laplacian matrix to achieve high-frequency graph attention, which is fused with features learned from self-attention to enhance the high-frequency feature representation. Given the uncertainty in pedestrian feature learning induced by occlusion makes it challenging to obtain reliable and stable pedestrian features, we propose a probability distribution learning module. This module establishes a memory bank to build Gaussian distributions for each pedestrian identity and the entropy is introduced as a loss function to encourage the model to generate more deterministic and relatively independent probability distributions, thereby enhancing the discriminative ability of the model across different pedestrian identities. The high-frequency enhancement module provides a solid foundation for the probability distribution learning module, alleviating uncertainty caused by pedestrian images themselves. Experimental results on occluded and holistic person re-identification datasets demonstrate the superiority of the proposed method.

Unsupervised person re-identification (ReID) aims to learn identity-discriminative representations without manual annotations, which is challenging due to noisy pseudo labels, background clutter, and large appearance variations. Recent studies have shown that exploiting fine-grained local cues is crucial for improving robustness in unsupervised ReID. In this context, random masking has emerged as a simple and annotation-free way to encourage the model to focus on informative regions. However, existing masking-based unsupervised ReID methods still suffer from two limitations: (1) Underused masked views: masked views are treated as degraded auxiliaries rather than exploited as fine-grained supervisory signals; (2) Weak cross-view alignment: feature alignment is restricted to mini-batch pairs, lacking explicit global alignment between masked and unmasked views across clusters. To address these issues, we propose the Mask-guided Asymmetric Contrastive and Semantic Alignment (ACSA) framework. Specifically, we introduce an Asymmetric Contrastive Learning (ACL) module with a dual-memory mechanism to separately encode masked and unmasked features, allowing masked views to serve as informative and discriminative supervision. In parallel, a Semantic Alignment Learning (SAL) module conducts multi-granularity distribution alignment by aligning both cluster-level prototypes and randomly sampled instance-level features, thereby preserving semantic consistency and intra-cluster diversity. Furthermore, to provide more reliable semantic anchors for SAL under noisy pseudo labels, we introduce a Progressive Refinement Module (PRM), which refines prototypes and features via exponential moving averaging for more stable semantic alignment. Extensive experiments validate the superiority of our method, even outperforming certain supervised counterparts. Code is available at https://github.com/Trangle12/ACSA.

Unsupervised person re-identification (USL-ReID) typically relies on clustering to generate pseudo-labels, but significant cross-view appearance variations often cause images of the same identity to be split into different clusters. Training on such noisy pseudo-labels severely degrades the learned representations. Therefore, learning robust view-invariant features is paramount. Data augmentation provides a direct way to enhance invariance, yet its trade-offs in USL-ReID remain under-explored: weak augmentations usually preserve identity semantics but lack diversity, whereas strong augmentations provide richer appearance diversity at the cost of partially corrupting identity-consistent semantic cues. To address this challenge, we propose Invariant Representation learning with Progressive Prototype Refinement (IRPP), a unified framework that learns invariant and discriminative features from noisy pseudo-labels. IRPP consists of three synergistic components. First, an Augmented Dual-Contrastive Learning (ADCL) module performs dataset-level prototype-guided invariant learning by contrasting weakly and strongly augmented views against cluster-derived prototypes. Second, an Alignment and Uniformity Learning (AUL) module regularizes the mini-batch-level weak-strong feature geometry, leading to more stable feature distributions under data augmentation. Third, a Progressive Prototype Refinement (PPR) mechanism progressively optimizes cluster centroids into cleaner prototypes, thereby mitigating the influence of noisy pseudo-labels and further strengthening invariant representation learning. This closed-loop design enables prototype-guided contrastive learning, weak-strong regularization, and prototype refinement to mutually reinforce each other. Extensive experiments on standard USL-ReID benchmarks demonstrate that IRPP achieves state-of-the-art performance with a simple and efficient training pipeline. Code is available at https://github.com/Trangle12/IRPP.

Unsupervised Domain Adaptation (UDA) person search aims to adapt models trained on labeled source data to unlabeled target domains. Existing approaches typically rely on clustering-based proxy learning, but their performance is often undermined by unreliable pseudo-supervision. This unreliability mainly stems from two challenges: (i) spectral shift bias, where low- and high-frequency components behave differently under domain shifts but are rarely considered, degrading feature stability; and (ii) static proxy updates, which make clustering proxies highly sensitive to noise and less adaptable to domain shifts. To address these challenges, we propose the Reliable Pseudo-supervision in UDA Person Search (RPPS) framework. At the feature level, a Dual-branch Wavelet Enhancement Module (DWEM) embedded in the backbone applies discrete wavelet transform (DWT) to decompose features into low- and high-frequency components, followed by differentiated enhancements that improve cross-domain robustness and discriminability. At the proxy level, a Dynamic Confidence-weighted Clustering Proxy (DCCP) employs confidence-guided initialization and a two-stage online-offline update strategy to stabilize proxy optimization and suppress proxy noise. Extensive experiments on the CUHK-SYSU and PRW benchmarks demonstrate that RPPS achieves state-of-the-art performance and strong robustness, underscoring the importance of enhancing pseudo-supervision reliability in UDA person search. Our code is accessible at https://github.com/zqx951102/RPPS.

Decoding visual information from human brain activity has seen remarkable advancements in recent research. However, the diversity in cortical parcellation and fMRI patterns across individuals has prompted the development of deep learning models tailored to each subject. The personalization limits the broader applicability of brain visual decoding in real-world scenarios. To address this issue, we introduce Wills Aligner, a novel approach designed to achieve multi-subject collaborative brain visual decoding. Wills Aligner begins by aligning the fMRI data from different subjects at the anatomical level. It then employs delicate mixture-of-brain-expert adapters and a meta-learning strategy to account for individual fMRI pattern differences. Additionally, Wills Aligner leverages the semantic relation of visual stimuli to guide the learning of inter-subject commonality, enabling visual decoding for each subject to draw insights from other subjects' data. We rigorously evaluate our Wills Aligner across various visual decoding tasks, including classification, cross-modal retrieval, and image reconstruction. The experimental results demonstrate that Wills Aligner achieves promising performance.

Decoding natural visual scenes from brain activity has flourished, with extensive research in single-subject tasks and, however, less in cross-subject tasks. Reconstructing high-quality images in cross-subject tasks is a challenging problem due to profound individual differences between subjects and the scarcity of data annotation. In this work, we proposed MindTuner for cross-subject visual decoding, which achieves high-quality and rich semantic reconstructions using only 1 hour of fMRI training data benefiting from the phenomena of visual fingerprint in the human visual system and a novel fMRI-to-text alignment paradigm. Firstly, we pre-train a multi-subject model among 7 subjects and fine-tune it with scarce data on new subjects, where LoRAs with Skip-LoRAs are utilized to learn the visual fingerprint. Then, we take the image modality as the intermediate pivot modality to achieve fMRI-to-text alignment, which achieves impressive fMRI-to-text retrieval performance and corrects fMRI-to-image reconstruction with fine-tuned semantics. The results of both qualitative and quantitative analyses demonstrate that MindTuner surpasses state-of-the-art cross-subject visual decoding models on the Natural Scenes Dataset (NSD), whether using training data of 1 hour or 40 hours.

Early exiting is an effective paradigm for improving the inference efficiency of pre-trained language models (PLMs) by dynamically adjusting the number of executed layers for each sample. However, in most existing works, easy and hard samples are treated equally by each classifier during training, which neglects the test-time early exiting behavior, leading to inconsistency between training and testing. Although some methods have tackled this issue under a fixed speed-up ratio, the challenge of flexibly adjusting the speed-up ratio while maintaining consistency between training and testing is still under-explored. To bridge the gap, we propose a novel Consistency-Oriented Signal-based Early Exiting (COSEE) framework, which leverages a calibrated sample weighting mechanism to enable each classifier to emphasize the samples that are more likely to exit at that classifier under various acceleration scenarios. Extensive experiments on the GLUE benchmark demonstrate the effectiveness of our COSEE across multiple exiting signals and backbones, yielding a better trade-off between performance and efficiency.

Concept-selective regions within the human cerebral cortex exhibit significant activation in response to specific visual stimuli associated with particular concepts. Precisely localizing these regions stands as a crucial long-term goal in neuroscience to grasp essential brain functions and mechanisms. Conventional experiment-driven approaches hinge on manually constructed visual stimulus collections and corresponding brain activity recordings, constraining the support and coverage of concept localization. Additionally, these stimuli often consist of concept objects in unnatural contexts and are potentially biased by subjective preferences, thus prompting concerns about the validity and generalizability of the identified regions. To address these limitations, we propose a data-driven exploration approach. By synthesizing extensive brain activity recordings, we statistically localize various concept-selective regions. Our proposed MindSimulator leverages advanced generative technologies to learn the probability distribution of brain activity conditioned on concept-oriented visual stimuli. This enables the creation of simulated brain recordings that reflect real neural response patterns. Using the synthetic recordings, we successfully localize several well-studied concept-selective regions and validate them against empirical findings, achieving promising prediction accuracy. The feasibility opens avenues for exploring novel concept-selective regions and provides prior hypotheses for future neuroscience research.

Early exiting has demonstrated great potential in accelerating the inference of pre-trained language models (PLMs) by enabling easy samples to exit at shallow layers, eliminating the need for executing deeper layers. However, existing early exiting methods primarily rely on class-relevant logits to formulate their exiting signals for estimating prediction certainty, neglecting the detrimental influence of class-irrelevant information in the features on prediction certainty. This leads to an overestimation of prediction certainty, causing premature exiting of samples with incorrect early predictions. To remedy this, we define an NSP score to estimate prediction certainty by considering the proportion of class-irrelevant information in the features. On this basis, we propose a novel early exiting method based on the Certainty-Aware Probability (CAP) score, which integrates insights from both logits and the NSP score to enhance prediction certainty estimation, thus enabling more reliable exiting decisions. The experimental results on the GLUE benchmark show that our method can achieve an average speed-up ratio of 2.19× across all tasks with negligible performance degradation, surpassing the state-of-the-art (SOTA) ConsistentEE by 28%, yielding a better trade-off between task performance and inference efficiency. The code is available at https://github.com/He-Jianing/NSP.git.

Current text-to-image diffusion generation typically employs complete-text conditioning. Due to the intricate syntax, diffusion transformers (DiTs) inherently suffer from a comprehension defect of complete-text captions. One-fly complete-text input either overlooks critical semantic details or causes semantic confusion by simultaneously modeling diverse semantic primitive types. To mitigate this defect of DiTs, we propose a novel split-text conditioning framework named DiT-ST. This framework converts a complete-text caption into a split-text caption, a collection of simplified sentences, to explicitly express various semantic primitives and their interconnections. The split-text caption is then injected into different denoising stages of DiT-ST in a hierarchical and incremental manner. Specifically, DiT-ST leverages Large Language Models to parse captions, extracting diverse primitives and hierarchically sorting out and constructing these primitives into a split-text input. Moreover, we partition the diffusion denoising process according to its differential sensitivities to diverse semantic primitive types and determine the appropriate timesteps to incrementally inject tokens of diverse semantic primitive types into input tokens via cross-attention. In this way, DiT-ST enhances the representation learning of specific semantic primitive types across different stages. Extensive experiments validate the effectiveness of our proposed DiT-ST in mitigating the complete-text comprehension defect. Datasets and models are available.

Pretrained language models have been widely applied in question-answering tasks. To achieve better generalization on unseen datasets, several previous studies often trained a single model on multiple datasets. However, owing to the noise and distribution differences among various datasets, the model tends to excessively adjust its weights during training on multiple datasets. This deviation from the initial pretraining state results in excellent performance on specific data but overfitting on other datasets. Ultimately, the model loses its ability to generalize to new data. In this paper, from the perspective of imposing constraints on model weights, we propose a novel fine-tuning method, binary mask tuning (BMT). We employ a carefully designed binary mask vector that is closely related to the data distribution to mask the gradient generated by backpropagation , achieving precise alignment of the subspace parameters required to fit the data from a huge parameter space. This approach aimed to enhance the adaptability of the model to the data distribution and improve parameter efficiency via more targeted fine-tuning. Our experiments demonstrate that BMT is not only effective in mitigating the tendency of the model to excessively adjust its weights but also in better capturing cross-dataset regularities and dataset-specific attributes in question-answering tasks across different datasets.

Graph neural networks (GNNs) have recently gained attention for capturing complex relations, prompting researchers to explore their potential in document classification. Existing studies serving this purpose fall into two directions: inductive learning focusing on personalized context relations within documents and transductive learning targeting the global distribution relations among documents in a corpus. Both directions extract distinct types of beneficial structural information and yield encouraging outcomes. However, due to the incompatibility of underlying graph structures and learning settings, developing an enhanced model that effectively integrates local and global relational learning within existing frameworks is challenging. To address this issue, we propose a new GNN-based document representation learning framework that incorporates multiview global graphs at both the word and document levels, focusing on learning the diverse global distribution information of texts at different granularities. Additionally, a contextual encoder derives the initial representations of document nodes from the updated representations of word nodes, integrating personalized context relations into document representations during this process. Finally, we tailor a node representation learning strategy for the multiview global graphs, called the multiview graph sampling and updating module, which allows our framework to operate efficiently during training without being constrained by the scale of the global graph. Experiments indicate that our framework generally enhances performance by integrating both global and local relational learning. When combined with large-scale language models, our framework achieves state-of-the-art results for GNN-based models across multiple datasets.

Multigranularity data analysis has recently become an active research topic in the intelligent computing and data mining fields. Feature selection via multigranularity data analysis is an effective tool for characterizing hierarchical data and enhancing the accuracy of the results. Although the multigranularity data analysis method has been widely adopted for feature selection, existing studies still present one prevalent disadvantage: multigranularity data analysis mostly focuses on information presented at a single granularity while ignoring the hierarchical structure of multigranularity data, which is contrary to the nature of multigranularity. Hence, this article proposes a multigranularity data analysis with a zentropy uncertainty measure for efficient and robust feature selection. Specifically, a consistent degree is first introduced to obtain optimal granularity combinations and establish an efficient neighborhood model for multigranularity information processing. Then, a novel and robust uncertainty measure is developed by integrating the multigranularity information, namely the zentropy-based measure. Considering its accuracy among uncertainty measures, two important measures are further designed and applied to feature selection. Extensive experiments demonstrate that the proposed method can achieve better robustness and classification performance than other state-of-the-art methods.

In order to process multiview multilabel, multilabel, and multiview data, current learning algorithms are designed on the basis of data characteristics, correlations, etc. While these algorithms cannot express correlations among different features, instances, labels in within-view, cross-view, and consensus-view representations self-adaptively and relative accurately. To this end, this study takes the classical multiple correlations-based model as the basis and explores some laws of self-adaptive change for those correlations in multiple representations. The proposed algorithm is called multiple self-adaptive correlation-based multiview multilabel learning (MuSC-MVML). Extensive experiments on 38 datasets demonstrate the superiority of MuSC-MVML and some conclusions are addressed. 1) MuSC-MVML outperforms most compared algorithms in statistical in terms of AUC and its performance is also stable; 2) the computational cost of MuSC-MVML is moderate and on most datasets, MuSC-MVML has a relatively fast convergence; and 3) introducing some laws of self-adaptive change for those correlations can improve the ability of MuSC-MVML to process multiview multilabel datasets effectively and express correlations in multiple representations better. Furthermore, this study explains the reason that why we use alternating optimization strategy to optimize the model of MuSC-MVML and provides some suggestions that how to modify the model of MuSC-MVML to process incomplete multiview multilabel datasets with noise.

The twin support vector machine (TWSVM) classifier and its fuzzy variant fuzzy twin support vector machine (FTSVM) have received considerable attention due to their low computational complexity. However, their performance often deteriorates when the input data is affected by noise. To overcome this limitation, this study leverages the robustness of granular-ball computing (GBC) against noise to develop more effective classification models by integrating GBC with TWSVM and FTSVM. First, we introduce the granular-ball TWSVM (GBTWSVM) classifier, which incorporates GBC with the TWSVM framework. By replacing traditional point-wise inputs with granular-ball representations, we derive a pair of nonparallel hyperplanes for the GBTWSVM classifier by solving a quadratic programming problem. Afterwards, we develop the granular-ball FTSVM (GBFTSVM) classifier, where the membership and nonmembership functions of granular-balls are defined using Pythagorean fuzzy sets, enabling a more nuanced differentiation of the contributions of granular-balls from distinct regions within the input space. By incorporating these functions into the FTSVM framework, we derive a pair of nonparallel hyperplanes for the GBFTSVM classifier through the solution of a quadratic programming problem. Finally, we present algorithms for the GBTWSVM and GBFTSVM classifiers and evaluate their performance on 21 benchmark datasets. Experimental results demonstrate the superior scalability, computational efficiency, and robustness of the proposed classifiers in pattern recognition, highlighting their potential as advanced tools for noise-tolerant classification.

The three-way approximation of fuzzy sets represents membership values using a three-valued set$\lbrace \mathbf{1}, \mathbf{m}, \mathbf{0}\rbrace$, where1indicates total belongingness,0total non-belongingness, andman intermediate state. This approach elevates values of membership function above a threshold$\alpha$to1, reduces those below$\beta$to0, and assigns the remaining ones to an intermediate valuem. A key challenge lies in determining the thresholds$\alpha$and$\beta$and selecting the value ofm, as existing models often lack analytical solutions and fail to fully explore the relationship betweenmand membership structures. This study introduces an asymmetric three-way approximation model for fuzzy sets, removing the constraint$\alpha + \beta = 1$. An analytical formula is derived for the thresholds$(\alpha, \beta)$by minimizing information loss, and the relationship betweenmand membership structures is thoroughly examined. An adaptive optimizer is proposed to learn the approximate optimal value ofmby minimizing the information loss. The experimental results show that information loss decreases initially before increasing asmgrows. Besides, our model achieves the best classification across most datasets.

The feature selection based on the rough set (RS) theory has been an active research topic in data mining and knowledge discovery. Fuzzy RSs (FRSs), an efficient tool to process the inconsistency between features and decisions, have attracted attention to the problems of feature selection. However, most FRSs-based feature selection methods pay much attention to the approximation space while ignoring the interaction between different levels. Note that the single-level feature selection method, depending on the boundary objects, is easily influenced by the noise data and cannot integrate multiple granular levels to evaluate features accurately. Therefore, this article proposes an efficient and robust feature selection approach based on the neighborhood-aware model and zentropy measure. Specifically, we first define a neighborhood-aware FRS (NAFRS) with weighted fuzzy relation to improve the antinoise ability of FRSs. Then, we propose a fuzzy granule zentropy (FGZE) measure based on zentropy by analyzing the granular level relation in NAFRS. Moreover, a significance measure with FGZE is designed and applied to feature selection. Finally, the experimental results of our method on 22 datasets by comparing it with 12 representative feature selection methods demonstrate the antinoise and the classification ability of the proposed method.

Multimodal content, such as mixing text with images, presents significant challenges to rumor detection in social media. Existing multimodal rumor detection has focused on mixing tokens among spatial and sequential locations for unimodal representation or fusing clues of rumor veracity across modalities. However, they suffer from less discriminative unimodal representation and are vulnerable to intricate location dependencies in the time-consuming fusion of spatial and sequential tokens. This work makes the first attempt at multimodal rumor detection in the frequency domain, which efficiently transforms spatial features into the frequency spectrum and obtains highly discriminative spectrum features for multimodal representation and fusion. A novel Frequency Spectrum Representation and fUsion network (FSRU) with dual contrastive learning reveals the frequency spectrum is more effective for multimodal representation and fusion, extracting the informative components for rumor detection. FSRU involves three novel mechanisms: utilizing the Fourier transform to convert features in the spatial domain to the frequency domain, the unimodal spectrum compression, and the cross-modal spectrum co-selection module in the frequency domain. Substantial experiments show that FSRU achieves satisfactory multimodal rumor detection performance.

Contrastive Language-Image Pretraining (CLIP) has achieved remarkable success, leading to rapid advancements in multimodal studies. However, CLIP faces a notable challenge in terms of *inefficient data utilization*. It relies on a single contrastive supervision for each image-text pair during representation learning, disregarding a substantial amount of valuable information that could offer richer supervision. Additionally, the retention of non-informative tokens leads to increased computational demands and time costs, particularly in CLIP's ViT image encoder. To address these issues, we propose **M**ulti-Perspective **L**anguage-**I**mage **P**retraining (**MLIP**). In MLIP, we leverage the frequency transform's sensitivity to both high and low-frequency variations, which complements the spatial domain's sensitivity limited to low-frequency variations only. By incorporating frequency transforms and token-level alignment, we expand CILP's single supervision into multi-domain and multi-level supervision, enabling a more thorough exploration of informative image features. Additionally, we introduce a token merging method guided by comprehensive semantics from the frequency and spatial domains. This allows us to merge tokens to multi-granularity tokens with a controllable compression rate to accelerate CLIP. Extensive experiments validate the effectiveness of our design.

The limited data availability and the low signal-to-noise ratio of fMRI signals lead to the challenging task of fMRI-to-image retrieval. State-of-the-art MindEye remarkably improves fMRI-to-image retrieval performance by leveraging a large model, i.e., a 996M MLP Backbone per subject, to align fMRI embeddings to the final hidden layer of CLIP's Vision Transformer (ViT). However, significant individual variations exist among subjects, even under identical experimental setups, mandating the training of large subject-specific models. The substantial parameters pose significant challenges in deploying fMRI decoding on practical devices. To this end, we propose Lite-Mind, a lightweight, efficient, and robust brain representation learning paradigm based on Discrete Fourier Transform (DFT), which efficiently aligns fMRI voxels to fine-grained information of CLIP. We elaborately design a DFT backbone with Spectrum Compression and Frequency Projector modules to learn informative and robust voxel embeddings. Our experiments demonstrate that Lite-Mind achieves an impressive 94.6% fMRI-to-image retrieval accuracy on the NSD dataset for Subject 1, with 98.7% fewer parameters than MindEye. Lite-Mind is also proven to be able to be migrated to smaller fMRI datasets and establishes a new state-of-the-art for zero-shot classification on the GOD dataset.

Reconstruction of static visual stimuli from non-invasion brain activity fMRI achieves great success, owning to advanced deep learning models such as CLIP and Stable Diffusion. However, the research on fMRI-to-video reconstruction remains limited since decoding the spatiotemporal perception of continuous visual experiences is formidably challenging. We contend that the key to addressing these challenges lies in accurately decoding both high-level semantics and low-level perception flows, as perceived by the brain in response to video stimuli. To the end, we propose NeuroClips, an innovative framework to decode high-fidelity and smooth video from fMRI. NeuroClips utilizes a semantics reconstructor to reconstruct video keyframes, guiding semantic accuracy and consistency, and employs a perception reconstructor to capture low-level perceptual details, ensuring video smoothness. During inference, it adopts a pre-trained T2V diffusion model injected with both keyframes and low-level perception flows for video reconstruction. Evaluated on a publicly available fMRI-video dataset, NeuroClips achieves smooth high-fidelity video reconstruction of up to 6s at 8FPS, gaining significant improvements over state-of-the-art models in various metrics, e.g., a 128% improvement in SSIM and an 81% improvement in spatiotemporal metrics. Our project is available at https://github.com/gongzix/NeuroClips.

Synthetic aperture radar (SAR) is a vital tool for ship detection, as it acquires high-resolution remote sensing images when optical images cannot penetrate. However, two primary challenges confronting SAR ship detection are complex backgrounds with islands, clutter, and land, as well as diverse scales of ship targets, particularly small ones, leading to numerous missed detections and false alarms. To overcome these challenges, we propose a multi-granularity-aware network (MGA-Net). Specifically, we design a multi-granularity hybrid feature fusion module (MGHF2M) to extract more representative local detail and global semantic information, enhancing the model’s capability to represent ship features to adapt to complex backgrounds. In addition, we design a multi-granularity feature synergy enhancement module (MGFSEM), which uses depthwise separable convolutions with different kernel sizes to extract features at different granularities and retain the original features, significantly improving the model’s representation of ship features at different scales. Experimental results show that our MGA-Net achieves the highest mAP and F1-score, surpassing eight advanced methods on three public datasets.

Most of the existing multi-attribute decision-making models under multi-scale decision information systems are established by selecting the optimal scale or fusing multi-scale information into a single scale. These models will lose part of the decision information, resulting in inaccurate decision results. However, sequential three-way decision can not only process information hierarchically, but also provide delayed decision between acceptance and rejection. In addition, the irrational behavior of decision-makers will have a certain impact on the decision-making results. To this end, for multi-scale and diversity decision-making problems, this paper proposes a hierarchical sequential three-way multi-attribute decision-making method based on regret theory. Specifically, to represent this diversity, the multi-scale evaluation information table is converted into a digital evaluation value table through a fuzzy membership function. Second, based on the regret-rejoicing function of regret theory, the regret-rejoicing relation of alternatives in multi-scale information systems is established, which can be used to calculate conditional probability. Third, the relative loss functions based on regret theory are proposed by considering the psychological behaviors of decision-makers. Finally, the hierarchical sequential three-way multi-attribute decision-making method for solving the multi-scale decision-making problem is proposed. The stability and effectiveness of the proposed method are verified by the corresponding experiments and the comparative analysis of practical cases. The proposed method solves the fusion problem of multi-scale decision information and obtains flexible ranking results according to the risk factor.

Feature selection and entropy theory are two efficacious data analysis tools for investigating uncertainty information processing in artificial intelligence. The fruitful marriage of the two has been an active research topic in knowledge discovery. Currently, most feature selection methods via entropy theory mainly focus on the information measures at a single granular level. However, it ignores the interaction between granular levels, which leads to the poor stability and accuracy of related methods. Hence, this article proposes a novel zentropy-based uncertainty measure to design a feature selection method by exploiting the granular level structure in knowledge space. Subsequently, by analyzing the granular level structure in decision data, the zentropy-based uncertainty measure and its properties are designed and analyzed to depict the uncertainty knowledge from whole and internal. Moreover, two importance measures are defined to evaluate features based on the designed uncertainty measure, and then a corresponding feature selection algorithm is developed. Finally, some experiments are carried out on public datasets to demonstrate that the proposed method can achieve state-of-the-art performance among methods, especially regarding stability and classification accuracy.

Feature selection facilitates system processing, and it relies on knowledge granulation and uncertainty measurement. Focusing on ordered decision systems, the fuzzy dominance neighborhood (FDN) granulation and corresponding condition entropy have recently yielded an outstanding algorithm for feature selection, FDNCE-FS (fuzzy dominance neighborhood condition entropy-based feature selection). However, there is room for improvement. Accordingly, three-level improvements of knowledge granulation, information enrichment, and heterogeneity fusion are proposed here, and$2\times 2\times 2=8$heuristic algorithms of feature selection are systematically established. First, FDN granulation is improved to fuzzy dominance three-way neighborhood (FD3N) granulation through three-way decision on fuzzy dominance degrees, and FD3N rough sets are modeled to offer better dependency. Second, the FDN condition entropy is improved to FD3N condition entropy by reinforcing the interaction factor and class information, and corresponding measure systems are constructed. Third, FD3N dependency is fused with four types of condition entropy to produce four combined measures, and eight uncertainty measures hierarchically emerge due to the three-level improvements. Fourth, these systematic measures have granulation nonmonotonicity, and they enable heuristic algorithms for feature selection; thus, the current FDNCE-FS method is improved to seven new selection algorithms: FHN-FS, RHN-FS, RFHN-FS, HTWN-FS, FHTWN-FS, RHTWN-FS, and RFHTWN-FS. Finally, the relevant FD3N granulation, uncertainty measurement, and feature selection are validated by data-based experiments, and the seven novel algorithms are shown to outperform FDNCE-FS in terms of classification performance. This study provides new insights into uncertainty modeling, information fusion, and feature selection through granular computing and three-way decision.

In recent years, person Re-IDentification (ReID) has rapidly progressed with wide real-world applications but is also susceptible to various forms of attack, including proven vulnerability to adversarial attacks. In this paper, we focus on the backdoor attack on deep ReID models. Existing backdoor attack methods follow an all-to-one or all-to-all attack scenario, where all the target classes in the test set have already been seen in the training set. However, ReID is a much more complex fine-grained open-set recognition problem, where the identities in the test set are not contained in the training set. Thus, previous backdoor attack methods for classification are not applicable to ReID. To ameliorate this issue, we propose a novel backdoor attack on deep ReID under a new all-to-unknown scenario, called Dynamic Triggers Invisible Backdoor Attack (DT-IBA). Instead of learning fixed triggers for the target classes from the training set, DT-IBA can dynamically generate new triggers for any unknown identities. Specifically, an identity hashing network is proposed to first extract target identity information from a reference image, which is then injected into the benign images by image steganography. We extensively validate the effectiveness and stealthiness of the proposed attack on benchmark datasets and evaluate the effectiveness of several defense methods against our attack.

In this article, we point out that the essential differences between convolutional neural network (CNN)-based and transformer-based detectors, which cause worse performance of small object in transformer-based methods, are the gap between local information and global dependencies in feature extraction and propagation. To address these differences, we propose a new vision transformer, called Hybrid Network Transformer (Hyneter), after preexperiments that indicate the gap causes CNN-based and transformer-based methods to increase size-different objects results unevenly. Different from the divide-and-conquer strategy in previous methods, Hyneters consist of hybrid network backbone (HNB) and dual switching (DS) module, which integrate local information and global dependencies, and transfer them simultaneously. Based on the balance strategy, HNB extends the range of local information by embedding convolution layers into transformer blocks in parallel, and DS adjusts excessive reliance on global dependencies outside the patch. Ablation studies illustrate that Hyneters achieve the state-of-the-art performance by a large margin of$+2.1\sim 13.2 \text{AP}$on COCO, and$+3.1\sim 6.5 \text{mIoU}$on VisDrone with lighter model size and lower computational cost in object detection. Furthermore, Hyneters achieve the state-of-the-art results on multiple computer vision tasks, such as object detection ($60.1 \text{AP}$on COCO and$46.1 \text{AP}$on VisDrone), semantic segmentation ($54.3 \text{AP}$on ADE20K), and instance segmentation ($48.5 \text{AP}^{\text{mask}}$on COCO), and surpass previous best methods. The code will be publicly available later.

Person search aims to find specific persons from visual scenes, including two subtasks, pedestrian detection, and person reidentification. The dominant fashion in this area is end-to-end networks that focus on analyzing the foreground (i.e., pedestrian) while ignoring the background (i.e., scene) information. However, the scene information often offers useful clues for person search. For example, pedestrians normally appear on the road rather than the top of a tree, and pedestrians appearing at the same location are likely to have similar occlusions. The interplay between the pedestrians and scenes can potentially improve the performance. In this article, a novel scene-pedestrian graph (SPG) is proposed, which can explicitly model the interplay between the pedestrians and scenes. To polish the quality of pedestrian bounding boxes, we pioneer a strategy of using the high-quality pedestrian bounding box to guide the low-quality one in the same scene. In addition, we design a contextual and temporal graph matching algorithm to effectively utilize the contextual and temporal information present in the constructed SPG to improve the performance of pedestrian matching. Benefiting from the robustness on complex scenes, our model achieves promising performance over the state-of-the-art methods on two popular person search benchmarks, CUHK-SYSU and PRW.

Person re-identification (ReID) typically encounters varying degrees of occlusion in real-world scenarios. While previous methods have addressed this using handcrafted partitions or external cues, they often compromise semantic information or increase network complexity. In this paper, we propose a new method from a novel perspective, termed as OAT. Specifically, we first use a Transformer backbone with multiple class tokens for diverse pedestrian feature learning. Given that the self-attention mechanism in the Transformer solely focuses on low-level feature correlations, neglecting higher-order relations among different body parts or regions. Thus, we propose the Second-Order Attention (SOA) module to capture more comprehensive features. To address computational efficiency, we further derive approximation formulations for implementing second-order attention. Observing that the importance of semantics associated with different class tokens varies due to the uncertainty of the location and size of occlusion, we propose the Entropy Guided Fusion (EGF) module for multiple class tokens. By conducting uncertainty analysis on each class token, higher weights are assigned to those with lower information entropy, while lower weights are assigned to class tokens with higher entropy. The dynamic weight adjustment can mitigate the impact of occlusion-induced uncertainty on feature learning, thereby facilitating the acquisition of discriminative class token representations. Extensive experiments have been conducted on occluded and holistic person re-identification datasets, which demonstrate the effectiveness of our proposed method.

Prompt learning stands out as one of the most efficient approaches for adapting powerful vision-language foundational models like CLIP to downstream datasets by tuning learnable prompt vectors with very few samples. However, despite its success in achieving remarkable performance on in-domain data, prompt learning still faces the significant challenge of effectively generalizing to novel classes and domains. Some existing methods address this concern by dynamically generating distinct prompts for different domains. Yet, they overlook the inherent potential of prompts to generalize across unseen domains. To address these limitations, our study introduces an innovative prompt learning paradigm, called MetaPrompt, aiming to directly learn domain invariant prompt in few-shot scenarios. To facilitate learning prompts for image and text inputs independently, we present a dual-modality prompt tuning network comprising two pairs of coupled encoders. Our study centers on an alternate episodic training algorithm to enrich the generalization capacity of the learned prompts. In contrast to traditional episodic training algorithms, our approach incorporates both in-domain updates and domain-split updates in a batch-wise manner. For in-domain updates, we introduce a novel asymmetric contrastive learning paradigm, where representations from the pre-trained encoder assume supervision to regularize prompts from the prompted encoder. To enhance performance on out-of-domain distribution, we propose a domain-split optimization on visual prompts for cross-domain tasks or textual prompts for cross-class tasks during domain-split updates. Extensive experiments across 11 datasets for base-to-new generalization and 4 datasets for domain generalization exhibit favorable performance. Compared with the state-of-the-art method, MetaPrompt achieves an absolute gain of 1.02% on the overall harmonic mean in base-to-new generalization and consistently demonstrates superiority over all benchmarks in domain generalization.

Knowledge discovery of heterogeneous data is an active topic in knowledge engineering. Feature selection for heterogeneous data is an important part of effective data analysis. Although there have been many attempts to study the feature selection for heterogeneous data, there are still some challenges, such as the unbalanced problem between the stability and validity of the designed model. Hence, this paper focuses on how to design an effective and robust heterogeneous feature selection method, namely a zentropy-based uncertainty measure for heterogeneous feature selection(Ze-HFS). Different from other entropy-based uncertainty measures, the proposed method does not consider single-level information measures but systematically analyzes and integrates the information between different granular levels, which has an obvious advantage in the study of heterogeneous data knowledge discovery. Specifically, a heterogeneous distance metric is first introduced to construct heterogeneous neighborhood granules and heterogeneous neighborhood rough sets(HNRS). Then, the zentropy-based uncertainty measure is developed by analyzing the granular level structure in the HNRS model. Finally, two significant measures based on the above research are designed for heterogeneous feature selection. Compared with other state-of-the-art methods, the experimental results on 18 public datasets demonstrate the robustness and effectiveness of the proposed method.

Outliers carry significant information to reflect an anomaly mechanism, so outlier detection facilitates relevant data mining. In terms of outlier detection, the classical approaches from distances apply to numerical data rather than nominal data, while the recent methods on basic rough sets deal with nominal data rather than numerical data. Aiming at wide outlier detection on numerical, nominal, and hybrid data, this paper investigates three-way neighborhood characteristic regions and corresponding fusion measurement to advance outlier detection. First, neighborhood rough sets are deepened via three-way decision, so they derive three-way neighborhood structures on model boundaries, inner regions, and characteristic regions. Second, the three-way neighborhood characteristic regions motivate the information fusion and weight measurement regarding all features, and thus, a multiple neighborhood outlier factor emerges to establish a new method of outlier detection; furthermore, a relevant outlier detection algorithm (called 3WNCROD) is designed to comprehensively process numerical, nominal, and mixed data. Finally, the 3WNCROD algorithm is experimentally validated, and it generally outperforms 13 contrast algorithms to perform better for outlier detection.

While person Re-identification (Re-ID) has progressed rapidly due to its wide real-world applications, it also causes severe risks of leaking personal information from training data. Thus, this paper focuses on quantifying this risk by membership inference (MI) attack. Most of the existing MI attack algorithms focus on classification models, while Re-ID follows a totally different training and inference paradigm. Re-ID is a fine-grained recognition task with complex feature embedding, and model outputs commonly used by existing MI like logits and losses are not accessible during inference. Since Re-ID focuses on modelling the relative relationship between image pairs instead of individual semantics, we conduct a formal and empirical analysis which validates that the distribution shift of the inter-sample similarity between training and test set is a critical criterion for Re-ID membership inference. As a result, we propose a novel membership inference attack method based on the inter-sample similarity distribution. Specifically, a set of anchor images are sampled to represent the similarity distribution conditioned on a target image, and a neural network with a novel anchor selection module is proposed to predict the membership of the target image. Our experiments validate the effectiveness of the proposed approach on both the Re-ID task and conventional classification task.

Speaker recognition achieved great progress recently, however, it is not easy or efficient to further improve its performance via traditional solutions: collecting more data and designing new neural networks. Aiming at the fundamental challenge of speech data, i.e. low information density, multimodal learning can mitigate this challenge by introducing richer and more discriminative information as input for identity recognition. Specifically, since the face image is more discriminative than the speech for identity recognition, we conduct multimodal learning by introducing a face recognition model (teacher) to transfer discriminative knowledge to a speaker recognition model (student) during training. However, this knowledge transfer via distillation is not trivial because the big domain gap between face and speech can easily lead to overfitting. In this work, we introduce a multimodal learning framework, VGSR (Vision-Guided Speaker Recognition). Specifically, we propose a MKD (Margin-based Knowledge Distillation) strategy for cross-modality distillation by introducing a loose constrain to align the teacher and student, greatly reducing overfitting. Our MKD strategy can easily adapt to various existing knowledge distillation methods. In addition, we propose a QAW (Quality-based Adaptive Weights) module to weight input samples via quantified data quality, leading to a robust model training. Experimental results on the VoxCeleb1 and CN-Celeb datasets show our proposed strategies can effectively improve the accuracy of speaker recognition by a margin of 10% ∼ 15%, and our methods are very robust to different noises.

In this paper, we point out that the essential differences between CNN-based and Transformer-based detectors, which cause worse performance of small object in Transformer-based methods, are the gap between local information and global dependencies in feature extraction and propagation. To address these differences, we propose a new vision Transformer, called Hybrid Network Transformer (Hyneter). Different from the divide and conquer strategy in previous methods, Hyneters consist of Hybrid Network Backbone (HNB) and Dual Switching module (DS), which integrate local information and global dependencies, and transfer them simultaneously. Based on the balance strategy, HNB extends the range of local information by embedding convolution layers into Transformer blocks, and DS adjusts excessive reliance on global dependencies outside the patch. Ablation studies illustrate that Hyneters surpass the state-of-the-art results on multiple vision tasks.

Vision Transformer (ViT) faces obstacles in wide application due to its huge computational cost. Almost all existing studies on compressing ViT adopt the manner of splitting an image with a single granularity, with very few exploration of splitting an image with multi-granularity. As we know, important information often randomly concentrate in few regions of an image, necessitating multi-granularity attention allocation to an image. Enlightened by this, we introduce the multi-granularity strategy to compress ViT, which is simple but effective. We propose a two-stage multi-granularity framework, MG-ViT, to balance ViT’s performance and computational cost. In single-granularity inference stage, an input image is split into a small number of patches for simple inference. If necessary, multi-granularity inference stage will be instigated, where the important patches are further subsplit into multi-finer-grained patches for subsequent inference. Moreover, prior studies on compression only for classification, while we extend the multi-granularity strategy to hierarchical ViT for downstream tasks such as detection and segmentation. Extensive experiments Prove the effectiveness of the multi-granularity strategy. For instance, on ImageNet, without any loss of performance, MG-ViT reduces 47\% FLOPs of LV-ViT-S and 56\% FLOPs of DeiT-S.

Few-shot object detection (FSOD) aims to transfer knowledge from base classes to novel classes, which receives widespread attention recently. The performance of current techniques is, however, limited by the poor classification ability and the improper features in the detection head. To circumvent this issue, we propose a Multi-level Feature Enhancement (MFE) model to improve the feature for classification from three different perspectives, including the spatial level, the task level and the regularization level. First, we revise the classifier’s input feature at the spatial level by using information from the regression head. Secondly, we separate the RoI-Align feature into two different feature distributions in order to improve features at the task level. Finally, taking into account the overfitting problem in FSOD, we design a simple but efficient regularization enhancement module to sample features into various distributions and enhance the regularization ability of classification. Extensive experiments show that our method achieves competitive results on PASCAL VOC datasets, and exceeds current state-of-the-art methods in all shot settings on challenging MS-COCO datasets.

Low-resolution text images are very commonplace in real life and their information is hard to be extracted by using existing text recognition methods only. Although this problem can be solved by introducing super-resolution (SR) techniques, most existing SR methods fail to process stroke regions and background regions of input text images distinctively. In this paper, we propose a text-specific super-resolution network named Text Enhanced Attention Network (TEAN) to solve this problem. First of all, we compensate for disadvantages of traditional thresholding mask operation proposed in Text Super-Resolution Network (TSRN) by utilizing deep-learning based semantic segmentation method to get correct masks as prior semantic information and propose a Text-Segmented-Contextual-Attention (TSCA) branch on the basis of them. Besides, we design an Orthogonal Contextual Attention Module (OCAM) working with TSCA to implicitly enhance stroke regions of LR images. Secondly, to effectively fuse shallow features and deep features of SR model, we propose a convolutional structure named Weight Balanced Fusion Module (WBFM) to improve traditional feature fusion methods of SR network. Finally, extensive experiments on TextZoom dataset demonstrate that the proposed network can improve the recognition accuracy of text images on existing text recognition models. Using TEAN to process low-resolution text images improves the recognition accuracy by 25.4% on CRNN, by 17.4% on ASTER, by 17.3% on MORAN, by 20.7% on NRTR, by 17.3% on SAR and by 15.9% on MASTER compared with directly recognizing them, which attains competitive performances against state-of-the-art methods. Furthermore, cross-dataset experiments on IC15_2077 demonstrate that TEAN is helpful for scene text recognition task, especially for low-resolution images even with the cross-domain issue.

With the continuous advancement of Remote Sensing (RS) technology, RS ship detection plays a crucial role in ensuring maritime safety and the oceanic economy, but it also faces various challenges. Most existing RS ship detection methods typically apply deblurring processing to all input images before using Feature Pyramid Network (FPN) to detect ships of different sizes. However, this indiscriminate operation may cause image quality degradation due to excessive deblurring. Moreover, FPN has limitations in fully utilizing multi-granularity features, which is particularly severe in RS ship detection tasks. These issues severely affect the accuracy of RS ship detection. To address these problems, this paper proposes an effective feature augmentation network, 3WM-AugNet, based on the three-way decisions (3WD) and multi-granularity feature learning for RS ship detection. It consists of two modules: a blurred classification and deblurring module (BCDM) and a multi-granularity feature augmentation module (MFAM). BCDM aims to combine 3WD and support vector machine (SVM) to design an image clarity classification algorithm and use the MT-RNN algorithm to process the blurry images classified, effectively avoiding excessive deblurring of clear images. MFAM is used to enhance the richness and robustness of feature representations for ships of different sizes by introducing the bottom-up feature fusion layer and designing an adaptive coordinate attention module. Experimental results on three commonly used datasets, FGSD2021, HRSC2016, and UCAS-AOD, show that our proposed 3WM-AugNet achieves state-of-the-art performance in RS ship detection.

Hierarchical classification is an important research hotspot in machine learning due to the widespread existence of data with hierarchical class structures. The existing sequential three-way decision models mainly constructed the hierarchical condition information granules via concept hierarchy tree to discuss the three probabilistic regions for flat classification. However, in real-world applications, one may face not only the tree-structured data with hierarchical condition attributes but also more often the multi-level data with hierarchical decision attribute (hierarchical class labels). How to obtain acceptable decisions under different levels of granularity is the most important issue within the multi-level and multi-view data. To this end, we construct a generalized hierarchical decision table and propose a generalized hierarchical multigranulation sequential three-way decision model by combining multi-granularity and sequential three-way decisions. Specifically, we first design a generalized hierarchical decision table using concept hierarchy trees of all conditional attributes and decision attribute, and explore some basic properties. Then we decompose and aggregate condition and decision granules under different levels of granularity, propose the optimistic and pessimistic generalized hierarchical multigranulation three-way decision models to update the three probabilistic regions for flat and hierarchical classification, and discuss the relationships between these two models. Finally, the experimental results demonstrate that the proposed models are more suitable for different applications. These models will provide a novel insight and enrich the development of multigranulation three-way decisions from the perspective of multi-level and multi-view.

Person search aims to locate and retrieve specific pedestrians in scene images, including two subtasks, pedestrian detection and person re-identification. Recently, triplet loss has been widely used in person re-identification, which effectively improves the pedestrian features embedding and achieves superior performance. However, forming triplet in the person search is not an easy task. Most of the existing end-to-end person search methods are based on Faster R-CNN. The training process of person re-identification part is affected by the detector. It is difficult to form pedestrian triplets within a limited batch size. Also, there are many pedestrian identities in the person search dataset, but each pedestrian identity only has a few samples. It is difficult to learn a robust pedestrian feature representation for person search. To resolve the problem discussed above, a novel Feature Compactness (FC) Loss for the person search is designed, which efficiently improves the inter-class discrimination and intra-class compactness of pedestrian features embedding without the need for positive or negative pairs. Besides, we propose a pedestrian attention module (PAM) to help the network focuses more on pedestrian information and suppresses irrelevant background information. Our method achieves comparable performance on two benchmarks, CUHK-SYSU and PRW, and achieves 91.96% of mAP and 93.34% of rank1 accuracy on CUHK-SYSU.

Person search aims to localize and identify specific pedestrians from numerous surveillance scene images. In this work, we focus on the noise in person search. We categorize the noise into scene-inherent noise and human-introduced noise. Scene-inherent noise comes from congestion, occlusion, and illumination changes. Human-introduced noise originates from the labeling process. For scene-inherent noise, we propose a novel context contrastive loss to take advantage of the latent contextual information from scene images. Features from context regions are utilized to construct contrastive pairs to constrain the feature discrimination among pedestrians in scene images while maintaining the feature consistency of the same identity. The network can thus learn to distinguish congested and overlapped pedestrians and more robust features can be obtained. For human-introduced noise, we propose a noise-discovery and noise-suppression training process for mislabeling robust person search. After the first training pass, the relation between feature prototypes of different identities is analyzed and the mislabeled pedestrians are discovered. During the second training pass, the label noise is suppressed to reduce the negative influence of mislabeled data. Experiments show that the proposed context-aware noise-robust (CANR) person search can achieve competitive performance. Further ablation studies confirm the effectiveness of CANR.

Abundant data with limited labeling are a widespread bottleneck in multilabel learning. Active learning (AL) is an effective solution to gradually enhance model robustness, however how to effectively extend instance selection criteria to multilabel case remains challenging. Considering the label specificity and label correlation, a granular batch mode-based ranking active model for the multilabel (GBRAML) is proposed. Taking a bottom-up view, three granulation operators are successively constructed to formulate three granular structures. In low-level granulation operator, auxiliary label is introduced to enhance the informativeness and representativeness of each label. The contribution of labels to the usefulness of instances is incorporated with pair-wise label correlation, and is considered in the middle-level granulation operator. The labeling priority is determined by ranking the scorings coming from high-level granulation operator. To alleviate the impact of skewed label correlation, we take a three-way strategy on fitness of representative label correlation, thus a three-way GBRAML model (TGBRAML) is devised. Extensive experiments on six multilabel benchmark demonstrate GBRAML gains 5.4% and 210.1% improvement on MicroF1 and Average Precision over state-of-the-art batch mode multilabel active learning. The effectiveness of three-way decisions in multilabel AL is also verified.

Person search aims at jointly solving Person Detection and Person Re-identification (re-ID). Existing works have designed end-to-end networks based on Faster R-CNN. However, due to the parallel structure of Faster R-CNN, the extracted features come from the low-quality proposals generated by the Region Proposal Network, rather than the detected high-quality bounding boxes. Person search is a fine-grained task and such inferior features will significantly reduce re-ID performance. To address this issue, we propose a Sequential End-to-end Network (SeqNet) to extract superior features. In SeqNet, detection and re-ID are considered as a progressive process and tackled with two sub-networks sequentially. In addition, we design a robust Context Bipartite Graph Matching (CBGM) algorithm to effectively employ context information as an important complementary cue for person matching. Extensive experiments on two widely used person search benchmarks, CUHK-SYSU and PRW, have shown that our method achieves state-of-the-art results. Also, our model runs at 11.5 fps on a single GPU and can be integrated into the existing end-to-end framework easily.

Person Re-identification (ReID) aims to retrieve the pedestrian with the same identity across different views. Existing studies mainly focus on improving accuracy, while ignoring their efficiency. Recently, several hash based methods have been proposed. Despite their improvement in efficiency, there still exists an unacceptable gap in accuracy between these methods and real-valued ones. Besides, few attempts have been made to simultaneously explicitly reduce redundancy and improve discrimination of hash codes, especially for short ones. Integrating Mutual learning may be a possible solution to reach this goal. However, it fails to utilize the complementary effect of teacher and student models. Additionally, it will degrade the performance of teacher models by treating two models equally. To address these issues, we propose a salience-guided iterative asymmetric mutual hashing (SIAMH) to achieve high-quality hash code generation and fast feature extraction. Specifically, a salience-guided self-distillation branch (SSB) is proposed to enable SIAMH to generate hash codes based on salience regions, thus explicitly reducing the redundancy between codes. Moreover, a novel iterative asymmetric mutual training strategy (IAMT) is proposed to alleviate drawbacks of common mutual learning, which can continuously refine the discriminative regions for SSB and extract regularized dark knowledge for two models as well. Extensive experiment results on five widely used datasets demonstrate the superiority of the proposed method in efficiency and accuracy when compared with existing state-of-the-art hashing and real-valued approaches. The code is released at https://github.com/Vill-Lab/SIAMH.

In some real-world situations, Pythagorean fuzzy sets are more powerful and effective than intuitionistic fuzzy sets to describe vague and uncertain information, and there are many Pythagorean fuzzy information systems for conflicts in which attitudes of agents on issues are depicted by Pythagorean fuzzy numbers. In this paper, we first provide the concepts of positive, neutral, and negative alliances with two thresholds and employ examples to illustrate how to compute positive, neutral, and negative alliances in Pythagorean fuzzy information systems for conflicts. Then, we focus on three-way conflict analysis based on the Bayesian minimum risk theory and explore examples to show how to compute the positive, neutral, and negative alliances with a Pythagorean fuzzy loss function given by an expert. Finally, we study how to calculate positive, neutral, and negative alliances with group decision theory and take examples to demonstrate how to construct the positive, neutral, and negative alliances with a group of Pythagorean fuzzy loss functions given by more experts.

Neighborhoods form a set-level approximation of data distribution for learning tasks. Due to the advantages of data generalization and nonparametric property, neighborhood models have been widely used for data classification. However, the existing neighborhood-based classification methods rigidly assign a certain class label to each data instance and lack the strategies to handle the uncertain instances. The far-fetched certain classification of uncertain instances may suffer serious risks. To tackle this problem, in this article, we propose a novel shadowed set to construct shadowed neighborhoods for uncertain data classification. For the fuzzy-rough transformation in the proposed shadowed set, a step function is utilized to map fuzzy neighborhood memberships to the set of triple typical values {0, 1, 0.5} and thereby partition a neighborhood into certain regions and uncertain boundary (neighborhood shadow). The threshold parameter in the step function for constructing shadowed neighborhoods is optimized through minimizing the membership loss in the mapping of shadowed sets. Based on the constructed shadowed neighborhoods, we implement a three-way classification algorithm to distinguish data instances into certain classes and uncertain case. Experiments validate the proposed three-way classification method with shadowed neighborhoods is effective in handling uncertain data and reducing the classification risk.

Person re-identification (Re-ID) is a challenging task in the field of computer vision and focuses on matching people across images from different cameras. The extraction of robust feature representations from pedestrian images through CNNs with a single deterministic pooling operation is problematic as the features in real pedestrian images are complex and diverse. To address this problem, we propose a novel center-triplet (CT) model that combines the learning of robust feature representation and the optimization of metric loss function. Firstly, we design a fusion feature learning network (FFLN) with a novel fusion strategy consisting of max pooling and average pooling. Instead of adopting a single deterministic pooling operation, the FFLN combines two pooling operations that can learn high response values, bright features, and low response values, discriminative features simultaneously. Our model obtains more discriminative fusion features by adaptively learning the weights of the features learned by the corresponding pooling operations. In addition, we design a hard mining center-triplet loss (HCTL), a novel improved triplet loss, which effectively optimizes the intra/inter-class distance and reduces the cost of computing and mining hard training samples simultaneously, thereby enhancing the learning of robust feature representation. Finally, we proved our method can learn robust and discriminative feature representations for complex pedestrian images in real scenes. The experimental results also illustrate that our method achieves an 81.8% mAP and a 93.8% rank-1 accuracy on Market1501, a 68.2% mAP and an 83.3% rank-1 accuracy on DukeMTMC-ReID, and a 43.6% mAP and a 74.3% rank-1 accuracy on MSMT17, outperforming most state-of-the-art methods and achieving better performance for person re-identification.

Ridge regression (RR) and its extended versions are widely used as an effective feature extraction method in pattern recognition. However, the RR-based methods are sensitive to the variations of data and can learn only limited number of projections for feature extraction and recognition. To address these problems, we propose a new method called robust discriminant regression (RDR) for feature extraction. In order to enhance the robustness, the L2,1-norm is used as the basic metric in the proposed RDR. The designed robust objective function in regression form can be solved by an iterative algorithm containing an eigenfunction, through which the optimal orthogonal projections of RDR can be obtained by eigen decomposition. The convergence analysis and computational complexity are presented. In addition, we also explore the intrinsic connections and differences between the RDR and some previous methods. Experiments on some well-known databases show that RDR is superior to the classical and very recent proposed methods reported in the literature, no matter the L2-norm or the L2,1-norm-based regression methods. The code of this paper can be downloaded from http://www.scholat.com/laizhihui.

This paper presents a novel image feature representation method, called multi-channel micro-structure difference descriptor (MCMSDD) for image retrieval. With the local feature extraction from a micro-structure and MAX operator, MCMSDD integrates the advantages of multi-channel local binary encoding and color difference histogram , which are the fusion of color, texture and spatial distribution information. Although it extracts feature from full color image, the dimension of the feature vector is relatively low without learning and segmentation. To improve the performance of retrieval, a simple re-ranking algorithm is employed. Finally, the proposed MCMSDD is extensively tested on Corel-2K and Washington datasets, and the experimental results show that the proposed MCMSDD is more effective than the state-of-the-art.

Lane detection based on computer vision is a key technology of Automatic Drive System for intelligent vehicles. In this paper, we propose a real-time and efficient lane detection algorithm that can detect lanes appearing in urban streets and highway roads under complex background. In order to enhance lane boundary information and to be suitable for various light conditions, we adopt canny algorithm for edge detection to get good feature points. We use the generalized curve lane parameter model, which can describe both straight and curved lanes. We propose an improved random sample consensus (RANSAC) algorithm combined with the least squares technique to estimate lane model parameters based on feature extraction. Experiments are conducted on both real road lane videos captured by Tongji University and Caltech Lane Datasets. The experimental results show that our algorithm is can meet the real time requirement and fit lane boundaries well in various challenging road conditions.

Principal curves arising as an essential construct in dimensionality reduction and data analysis have recently attracted much attention from theoretical as well as practical perspective. In many real-world situations, however, the efficiency of existing principal curves algorithms is often arguable, in particular when dealing with massive data owing to the associated high computational complexity. A certain drawback of these constructs stems from the fact that in several applications principal curves cannot fully capture some essential problem-oriented facets of the data dealing with width, aspect ratio, width change, etc. Information granulation is a powerful tool supporting processing and interpreting massive data. In this paper, invoking the underlying ideas of information granulation, we propose a granular principal curves approach, regarded as an extension of principal curves algorithms, to improve efficiency and achieve a sound accuracy-efficiency tradeoff. First, large amounts of numerical data are granulated into C intervals-information granules developed with the use of fuzzy C-means clustering and the two criteria of information granulation, which significantly reduce the amount of data to be processed at the later phase of the overall design. Granular principal curves are then constructed by determining the upper and the lower bounds of the interval data. Finally, we develop an objective function using the criteria of information confidence and specificity to evaluate the granular output formed by the principal curves. We also optimize the granular principal curves by adjusting the level of information granularity (the number of clusters), which is realized with the aid of the particle swarm optimization. A number of numeric studies completed for synthetic and real-world datasets provide a useful quantifiable insight into the effectiveness of the proposed algorithm.

Traditional document clustering approaches are usually based on the Bag of Words model, which is limited by its assumption of the independence among terms. Recent strategies have been proposed to capture the relation between terms based on statistical analysis, and they estimate the relation between terms purely by their co-occurrence across the documents. However, the implicit interactions with other link terms are overlooked, which leads to the discovery of incomplete information. This paper proposes a coupled term-term relation model for document representation, which considers both the intra-relation (i.e. co-occurrence of terms) and inter-relation (i.e. dependency of terms via link terms) between a pair of terms. The coupled relation for each pair of terms is further used to map a document onto a new feature space, which includes more semantic information. Substantial experiments verify that the document clustering incorporated with our proposed relation achieves a significant performance improvement compared to the state-of-the-art techniques.

Forensic electroencephalogram (EEG)-based lie detection has recently begun using the concealed information test (CIT) as a potentially more robust alternative to the classical comparative questions test. The main problem with using CIT is that it requires an objective and fast decision algorithm under the constraint of limited available information. In this study, we developed a simple and feasible hierarchical knowledge base construction and test method for efficient concealed information detection based on objective EEG measures. We describe how a hierarchical feature space was formed and which level of the feature space was sufficient to accurately predict concealed information from the raw EEG signal in a short time. A total of 11 subjects went through an autobiographical paradigm test. A high accuracy of 95.23% in recognizing concealed information with a single EEG electrode within about 20 seconds demonstrates effectiveness of the method.

A framework is proposed for studying a particular class of set-theoretic approaches to granular computing. A granule is a subset of a universal set, a granular structure is a family of subsets of the universal set, and relationship between granules i

It is difficult but crucial for minutiae extraction and pseudo minutiae deletion of low quality fingerprint images in auto fingerprint identification systems. Traditional methods based on thinning images or gray-level images are, however, susceptible to noise. Reference 14 indicated that principal curves based fingerprint minutiae extraction was feasible to overcome the drawback, but the extended polygonal line (EPL) principal curves algorithm used in the paper extracted the principal curves ineffectively. As the fingerprint data sets are usually large, the original EPL principal curves algorithm is time-consuming. Meanwhile, scattered fingerprint data lead to the deviation of fingerprint skeleton. In this paper, the algorithm is modified, and a fingerprint minutiae extraction and pseudo minutiae detection method based on principal curves is proposed. Experimental results show that the modified EPL principal curves algorithm outperforms the original EPL algorithm both in efficiency and quality, and the proposed minutiae extraction method outperforms the methods proposed by Miao under noise conditions.

Most of the business decisions are based on cost and benefit considerations. Data mining techniques that make it possible for the businesses to incorporate financial considerations will be moremeaningful to the decisionmakers. Decision theoretic framework has been helpful in providing a better understanding of classification models. This study describes a semi-supervised decision theoretic rough set model. The model is based on an extension of decision theoretic model proposed by Yao. The proposal is used to model financial cost/benefit scenarios for a promotional campaign in a real-world retail store.

Quality of clustering is an important issue in application of clustering techniques. Most traditional cluster validity indices are geometry-based cluster quality measures. This paper proposes a cluster validity index based on the decision-theoretic rough set model by considering various loss functions. Experiments with synthetic, standard, and real-world retail data show the usefulness of the proposed validity index for the evaluation of rough and crisp clustering. The measure is shown to help determine optimal number of clusters, as well as an important parameter called threshold in rough clustering. The experiments with a promotional campaign for the retail data illustrate the ability of the proposed measure to incorporate financial considerations in evaluating quality of a clustering scheme. This ability to deal with monetary values distinguishes the proposed decision-theoretic measure from other distance-based measures. The proposed validity index can also be extended for evaluating other clustering algorithms such as fuzzy clustering.

A great number of biclustering algorithms have been proposed for analyzing gene expression data. Many of them assume to find exclusive biclusters whose subsets of genes are co-regulated under subsets of conditions without intersection. This is not consistent with a general understanding of biological processes that many genes participate in multiple different processes. Therefore nonexclusive biclustering algorithms are required. In this paper we present a novel approach (ROB) to find potentially overlapping biclusters in the framework of generalized rough sets. Our scheme mainly consists of two phases. First, we generate a set of highly coherent seeds (original biclusters) based on two-way rough k-means clustering. And then, the seeds are iteratively adjusted (enlarged or degenerated) by adding or removing genes and conditions based on a proposed criterion. We illustrate the method on yeast gene expression data. The experiments demonstrate the effectiveness of this approach.

This paper introduces a novel approach to large scale content management based on the synergy and mashup of the semantic web and collaborative tagging technologies. We propose a generic conceptual architecture for community oriented semantic tagging and discusses the functionality and interplay of its key components. The approach has been applied to a real world application - an open online publishing system (OOPS) to enable individuals to achieve flexible online publishing, open accesses and effective discovery. A prototype OOPS system and underpinning technological infrastructure have been developed to demonstrate the approach.