Incomplete multi-view clustering (IMVC) aims to discover shared cluster structures from multi-view data with partial observations. The core challenges lie in accurately imputing missing views without introducing bias, while maintaining semantic consistency across views and compactness within clusters. To address these challenges, we propose DIMVC-HIA, a novel deep IMVC framework that integrates hierarchical imputation and alignment with four key components: (1) view-specific autoencoders for latent feature extraction, coupled with a view-shared clustering predictor to produce soft cluster assignments; (2) a hierarchical imputation module that first estimates missing cluster assignments based on cross-view contrastive similarity, and then reconstructs missing features using intra-view, intra-cluster statistics; (3) an energy-based semantic alignment module, which promotes intra-cluster compactness by minimizing energy variance around low-energy cluster anchors; and (4) a contrastive assignment alignment module, which enhances cross-view consistency and encourages confident, well-separated cluster predictions. Experiments on benchmarks demonstrate that our framework achieves superior performance under varying levels of missingness.

Heterogeneous Graph Neural Networks (HGNNs) excel in modeling complex, multi-typed relationships across diverse domains, yet their vulnerability to backdoor attacks remains unexplored. To address this gap, we conduct the first investigation into the susceptibility of HGNNs to existing graph backdoor attacks, revealing three critical issues: (1) high attack budget required for effective backdoor injection, (2) inefficient and unreliable backdoor activation, and (3) inaccurate attack effectiveness evaluation. To tackle these issues, we propose the Heterogeneous Graph Backdoor Attack (HGBA), the first backdoor attack specifically designed for HGNNs, introducing a novel relation-based trigger mechanism that establishes specific connections between a strategically selected trigger node and poisoned nodes via the backdoor metapath. HGBA achieves efficient and stealthy backdoor injection with minimal structural modifications and supports easy backdoor activation through two flexible strategies: Self-Node Attack and Indiscriminate Attack. Additionally, we improve the ASR measurement protocol, enabling a more accurate assessment of attack effectiveness. Extensive experiments demonstrate that HGBA far surpasses multiple state-of-the-art graph backdoor attacks in black-box settings, efficiently attacking HGNNs with low attack budgets. Ablation studies show that the strength of HBGA benefits from our trigger node selection method and backdoor metapath selection strategy. In addition, HGBA shows superior robustness against node feature perturbations and multiple types of existing graph backdoor defense mechanisms. Finally, extension experiments demonstrate that the relation-based trigger mechanism can effectively extend to tasks in homogeneous graph scenarios, thereby posing severe threats to broader security-critical domains.

Sequential recommendation aims to predict the next item that a user will interact with based on historical behavior sequences. In real-world scenarios, user-item interactions exhibit complex dependencies, which graph neural networks are well-suited to model by capturing high-order relationships between nodes. However, most existing graph-based sequential recommendation methods face two major challenges: 1) they often neglect positional information within sequences when constructing graphs; and 2) they suffer from noise introduced by accidental or unintended clicks. Recent advances in large language models (LLMs) offer a promising direction for mitigating these issues, due to their strong semantic understanding. However, directly leveraging LLMs may face task mismatch and excessive denoising may exacerbate the data sparsity. To this end, we propose an LLM-enhanced position-aware graph for sequential recommendation (LEPG4SR). Specifically, we design a position-aware item transition graph to model complex item relationships from a global perspective. We then utilize LLMs to extract semantic embeddings of item side information and filter out noisy data based on semantic similarity. To further combat data sparsity, we introduce a self-supervised learning strategy with a novel semantic perturbation-based data augmentation technique. Extensive experiments on three real-world datasets demonstrate that LEPG4SR can outperform the state-of-the-art sequential recommendation methods.

Incomplete multiview clustering (IMVC) faces significant challenges due to missing data and inherent view discrepancies. While deep neural networks offer powerful representation learning capabilities for IMVC, existing methods often overlook view diversity and force representations across views to be identical, leading to 1) biased representations with distorted topologies and 2) inaccurate imputation for missing data, ultimately degrading clustering performance. To address these issues, we propose prototype-graph transformer (PGFormer), a novel IMVC framework that integrates prototype assignments, rather than direct representations, to enhance clustering performance. PGFormer leverages view-specific encoders to extract features from available samples in each view, employs a PGFormer designed to refine node embeddings, and reconstructs available samples using these refined embeddings. For each view, PGFormer utilizes a graph convolutional network (GCN) to model node-to-node topologies and generate semantic prototypes from the node embeddings. These view-specific prototypes and embeddings are then refined through dual attention mechanisms: prototype-to-prototype (P2P) self-attention and prototype-to-node (P2N) cross-attention, enabling a thorough exploration of multilevel topological relationships within each view. To address missing data, the cross-prototype imputation (CPI) module leverages the weighted prototype assignments from different views to impute missing samples using refined intraview prototypes. Building on this, the cross-view alignment module calibrates prototype assignments to ensure consistent predictions across views. Extensive experiments demonstrate that PGFormer can achieve superior performance compared with the baselines.

Incomplete multi-view clustering (IMVC) deals with real-world scenarios where certain views are partially missing, posing significant challenges to effective clustering. Most existing IMVC approaches face a trade-off: imputation-free methods suffer from information bias and imbalance, while full-imputation methods risk introducing and propagating noise. To overcome these limitations, we propose Energy-Based Deep Incomplete Multi-View Clustering (Energy-DIMC), a novel selective-imputation framework that leverages energy-based models (EBMs) to guide reliable imputations and robust clustering. EBMs assess data compatibility by assigning lower energy to more coherent structures, effectively modeling complex inter-view and inter-sample dependencies. Inspired by EBMs, Energy-DIMC integrates four key components: 1) a view feature projector that learns view-specific features and projects them into a common feature space; 2) an energy-guided selective imputation module that identifies the most reliable source view for each view based on view energies, and performs feature imputation only when cross-view transfer is feasible, avoiding unreliable imputations; 3) an energy-based representation fusion module that aggregates observed and selectively imputed features across views via a view attention mechanism, generating view-coherent representations; 4) an energy-enhanced contrastive alignment module that enforces consistency between view-specific and view-coherent representations using dual-level energy signals to preserve true positives. Extensive experiments demonstrate that Energy-DIMC outperforms state-of-the-art IMVC methods across diverse missing-view scenarios. The code is available at https://github.com/sunway677/EnergyIMVC.

Few-shot class-incremental learning (FSCIL) aims to learn new classes incrementally with a limited number of samples per class. Foundation models combined with prompt tuning showcase robust generalization and zero-shot learning (ZSL) capabilities, endowing them with potential advantages in transfer capabilities for FSCIL. However, existing prompt tuning methods excel in optimizing for stationary datasets, diverging from the inherent sequential nature in the FSCIL paradigm. To address this issue, taking inspiration from the "fast and slow mechanism" of the complementary learning systems (CLSs) in the brain, we present fast- and slow-update prompt tuning FSCIL (FSPT-FSCIL), a brain-inspired prompt tuning method for transferring foundation models to the FSCIL task. We categorize the prompts into two groups: fast-update prompts and slow-update prompts, which are interactively trained through meta-learning. Fast-update prompts aim to learn new knowledge within a limited number of iterations, while slow-update prompts serve as meta-knowledge and aim to strike a balance between rapid learning and avoiding catastrophic forgetting. Through experiments on multiple benchmark tests, we demonstrate the effectiveness and superiority of FSPT-FSCIL. The code is available at https://github.com/qihangran/FSPT-FSCIL.

Most state-of-the-art convolutional neural networks (CNNs) are characterized by excessive parameterization, leading to a high computational burden. Tensor decomposition has emerged as a model reduction technique for compressing deep neural networks. Previous approaches have predominantly relied on either Tucker decomposition or Canonical Polyadic (CP) decomposition for CNNs. However, CP decomposition exhibits exceptional compression capabilities in comparison to Tucker decomposition, which results in a more pronounced accuracy loss. This article introduces an efficient model compression method, termed TEC-CNN, designed to achieve significant compression while preserving accuracy levels comparable to those of the original models. In TEC-CNN, convolutional layers are identified to obtain convolutional kernels by analyzing given models under the principles of low-rank tensor decomposition, and then, calculating the ranks of convolutional kernels. Furthermore, an efficient decomposition schema for the convolutional kernel is proposed with approximate kernel tensor for reducing parameters. Additionally, a novel format of a convolutional sequence is presented and constructed with a reduced number of parameters to replace the original convolutional layers. Finally, the effectiveness of TEC-CNN is assessed across a range of computer vision tasks. For instance, in CIFAR-100 classification, ResNet18 is compressed to 4.1 MB, while Unext, when applied to image segmentation using the International Skin Imaging Collaboration (ISIC) dataset, is reduced to 3.419 MB. When employed for fire object detection with Yolov7, TEC-CNN achieves a model size reduction of 71.6 MB. Comprehensive experimental results underscore that our approach achieves significant model compression while preserving model performance.

Exploring the structure information is crucial for data clustering task, particularly for the sceneries of incomplete multiview clustering (IMVC) when some views are missing. However, almost all of the existing graph-based IMVC methods either introduce the Laplacian constraint with fixed graphs or simply fuse the graphs of all views, which are vulnerable to the quality of the constructed graphs. To address this issue, we propose a new graph-based method, called confident local structure-aware incomplete multiview spectral clustering. Different from existing works, our method seeks to adaptively uncover the inherent similarity structure among the available instances in each view and learn the optimal consensus graph within a unified learning framework. Moreover, to mitigate the adverse effects of imbalance information across incomplete views and improve the quality of consensus graph, we further impose some adaptive weights on the consensus graph learning model w.r.t. each view and introduce some confident structure graphs to explore the most confident similarity information in the model. In contrast to existing works, our approach simultaneously takes into account the pairwise similarity information and neighbor group-based confident structure information. This dual consideration makes our method more effective in achieving the optimal consensus graph and delivering superior IMVC performance. Experimental results on several datasets demonstrate that our method effectively learns a high-quality and clustering-friendly graph from incomplete multiview data, and it outperforms many state-of-the-art IMVC methods in terms of clustering performance.

Few-Shot Class-Incremental Learning (FSCIL) aims to learn new classes incrementally with a limited number of samples per class. It faces issues of forgetting previously learned classes and overfitting on few-shot classes. An efficient strategy is to learn features that are discriminative in both base and incremental sessions. Current methods improve discriminability by manually designing inter-class margins based on empirical observations, which can be suboptimal. The emerging Neural Collapse (NC) theory provides a theoretically optimal inter-class margin for classification, serving as a basis for adaptively computing the margin. Yet, it is designed for closed, balanced data, not for sequential or few-shot imbalanced data. To address this gap, we propose a Meta-learning- and NC-based FSCIL method, MetaNC-FSCIL, to compute the optimal margin adaptively and maintain it at each incremental session. Specifically, we first compute the theoretically optimal margin based on the NC theory. Then we introduce a novel loss function to ensure that the loss value is minimized precisely when the inter-class margin reaches its theoretically best. Motivated by the intuition that “learn how to preserve the margin” matches the meta-learning’s goal of “learn how to learn”, we embed the loss function in base-session meta-training to preserve the margin for future meta-testing sessions. Experimental results demonstrate the effectiveness of MetaNC-FSCIL, achieving superior performance on multiple datasets. The code is available at https://github.com/qihangran/metaNC-FSCIL.

Large deep learning models are impressive, but they struggle when real-time data is not available. Few-shot class-incremental learning (FSCIL) poses a significant challenge for deep neural networks to learn new tasks from just a few labeled samples without forgetting the previously learned ones. This setup can easily leads to catastrophic forgetting and overfitting problems, severely affecting model performance. Studying FSCIL helps overcome deep learning model limitations on data volume and acquisition time, while improving practicality and adaptability of machine learning models. This paper provides a comprehensive survey on FSCIL. Unlike previous surveys, we aim to synthesize few-shot learning and incremental learning, focusing on introducing FSCIL from two perspectives, while reviewing over 30 theoretical research studies and more than 20 applied research studies. From the theoretical perspective, we provide a novel categorization approach that divides the field into five subcategories, including traditional machine learning methods, meta learning-based methods, feature and feature space-based methods, replay-based methods, and dynamic network structure-based methods. We also evaluate the performance of recent theoretical research on benchmark datasets of FSCIL. From the application perspective, FSCIL has achieved impressive achievements in various fields of computer vision such as image classification, object detection, and image segmentation, as well as in natural language processing and graph. We summarize the important applications. Finally, we point out potential future research directions, including applications, problem setups, and theory development. Overall, this paper offers a comprehensive analysis of the latest advances in FSCIL from a methodological, performance, and application perspective.

Recent development in computing power has resulted in performance improvements on holistic(none-occluded) person Re-Identification (ReID) tasks. Nevertheless, the precision of the recent research will diminish when a pedestrian is obstructed by obstacles. Within the realm of 2D space, the loss of information from obstructed objects continues to pose significant challenges in the context of person ReID. Person is a 3D non-grid object, and thus semantic representation learning in only 2D space limits the understanding of occluded person. In the present work, we propose a network based on 3D multi-view learning, allowing it to acquire geometric and shape details of an occluded pedestrian from 3D space. Simultaneously, it capitalizes on advancements in 2D-based networks to extract semantic representations from 3D multi-views. Specifically, the surface random selection strategy is proposed to convert images of 2D RGB into 3D multi-views. Using this strategy, we build four extensive 3D multi-view data collections for person ReID. After that, Pedestrian 3D Shape Understanding for Person Re-Identification via Multi-View Learning(MV-3DSReID), is proposed for identifying the person by learning person geometry and structure representation from the groups of multi-view images. In comparison to alternative data formats (e.g., 2D RGB, 3D point cloud), multi-view images complement each other’s detailed features of the 3D object by adjusting rendering viewpoints, thus facilitating a more comprehensive understanding of the person for both holistic and occluded ReID situations. Experiments on occluded and holistic ReID tasks demonstrate performance levels comparable to state-of-the-art methods, validating the effectiveness of our proposed approach in tackling challenges related to occlusion. The code is available at https://github.com/hangjiaqi1/MV-TransReID.

Emerging machine learning approaches provide effective solutions in a data-driven manner for cooperative spectrum sensing (CSS). Their success relies on large amounts of labeled training data to capture spectrum characteristics. However, data collection and annotation under dynamic spectrum environments are time-consuming and impractical. To this end, we propose a novel Domain Adaptation-based Detector for CSS (DADCSS) to improve the sensing adaptiveness and robustness under dynamic environments. It learns environment-invariant features, reduces the domain shift via the inter-domain and intraclass feature alignments, and determines the channel status by an adaptive detector trained by the aligned features. Simulation results demonstrate the effectiveness of our proposed method.

In existing graph-based multi-view clustering algorithms, consensus cluster structures are explored by constructing similarity graphs of multiple views and then fusing them into a unified superior graph. However, they overlook consensus information when learning each graph independently, resulting in the undesirable unified graph with biases. To this end, we proposed a framework named bipartite graph based multi-view clustering (BIGMC) in [1] to tackle this challenge. To summarize, the key idea of BIGMC is to employ a small number of uniform anchors to represent the consensus information across views. In this way, BIGMC creates a bipartite graph between data points and anchors for each view, which are then fused to generate a unified bipartite graph. The unified graph would in turn improve each view bipartite graph and the anchor set. Finally, the clusters are formed directly using the unified graph. In this extended abstract, we also summarize the effectiveness of BIGMC as shown in experimental results originally presented in [1].

Cross-lingual word alignment is the task for word translation between monolingual word embedding spaces of two different languages. Recent work is mostly based on supervised approaches, while their success relies on bilingual seed dictionaries derived from aligned data. The unsupervised adversarial approaches, which utilize generative adversarial networks (GANs) to map the global monolingual space to another space, can eliminate the need for aligned data. However, most GAN-based unsupervised approaches ignore the issues of mode collapse and gradient disappearance in GANs, leading to a training failure to converge. In addition, these approaches often fail to account for the low isomorphism between language pairs, which prevents capturing the non-linear relationship contained in cross-lingual embedding spaces. To address these issues, we propose a novel unsupervised unified framework with an adaptive training objective for the GANs' improvement (ATOGAN) and a local mapping (LM) strategy for exploring the non-linear relationship. We present ATOGAN to learn bi-directional global mapping using unaligned word embeddings, which integrates particle swarm optimization (PSO) to adaptively select the training objective for preventing mode collapse and gradient disappearance. Then, we design an LM strategy based on the guidance of dictionaries generated by trained ATOGAN to alleviate reliance on isomorphism assumption for purely linear mapping. Experimental results demonstrate the effectiveness of our proposed method for cross-lingual word alignment in low isomorphic embedding spaces (distant language pairs). Our code is available athttps://github.com/goFurtherLong/ATOGAN.

Incomplete multi-view clustering (IMC) aims to integrate the complementary information from incomplete views to improve clustering performance. Most existing IMC methods try to fill the incomplete views or directly learn a common representation based on matrix factorization or subspace learning. The former may introduce useless even noisy information especially for data with a large missing ratio. The latter relies on the initialization and ignores the geometric structure of data. To address these issues, we propose a novel Joint Partition and Graph (JPG) learning method for IMC. Specifically, JPG jointly constructs local incomplete graph matrices, generates incomplete base partition matrices, stretches them to produce a unified partition matrix, and employs it to learn a consensus graph matrix. By this means, we transform incomplete multi-view data into a unified partition space and obtain the consensus graph in a mutual reinforcement manner. Moreover, a partition fusion strategy can allocate a large weight to the stretched base partition that is close to the unified matrix. The objective function is optimized in an alternating optimization fashion. Experimental results on several benchmark datasets demonstrate the effectiveness and superiority of JPG than the state-of-the-art baselines

Unsupervised domain adaptation (DA) aims to perform classification tasks on the target domain by leveraging rich labeled data in the existing source domain. The key insight of DA is to reduce domain divergence by learning domain-invariant features or transferable instances. Despite its rapid development, there still exist several challenges to explore. At the feature level, aligning both domains only in a single way (i.e., geometrical or statistical) has limited ability to reduce the domain divergence. At the instance level, interfering instances often obstruct learning a discriminant subspace when performing the geometrical alignment. At the classifier level, only minimizing the empirical risk on the source domain may result in a negative transfer. To tackle these challenges, this article proposes a novel DA method, called discriminant geometrical and statistical alignment (DGSA). DGSA first aligns the geometrical structure of both domains by projecting original space into a Grassmann manifold, then matches the statistical distributions of both domains by minimizing their maximum mean discrepancy on the manifold. In the former step, DGSA only selects the density peaks to learn the Grassmann manifold and so to reduce the influences of interfering instances. In addition, DGSA exploits the high-confidence soft labels of target landmarks to learn a more discriminant manifold. In the latter step, a structural risk minimization (SRM) classifier is learned to match the distributions (both marginal and conditional) and predict the target labels at the same time. Extensive experiments on objection recognition and human activity recognition tasks demonstrate that DGSA can achieve better performance than the comparison methods.

Fault diagnosis that identifies the root of the abnormal status is of great importance to eliminate faults in the complex chemical processes. Many data-driven fault diagnosis models ignore different faults that occur with varied frequencies in chemical plants, and they need a complete retraining process with the arrival of new fault modes. In this article, a novel incremental imbalance modified convolutional neural network is proposed to solve the aforementioned issues. The proposed method employs an imbalance modified method to extract the valuable information from the imbalance data, and generate new samples. After that, a local hyperplane-based dynamic Relief is designed to reduce the dimension of the chemical data and simplify the complex learning process. Finally, for the arrival of new fault modes, the proposed method is prompted in an incremental hierarchical way. Unlike the traditional models that are trained on static data, the proposed method inherits the existing knowledge and updates itself to include new coming fault classes. The proposed method is utilized in a simulated process and a real industrial process. Experimental results illustrate that the proposed method is better than the existing methods and has significant robustness and reliability in chemical fault diagnosis.

For graph-based multi-view clustering, a critical issue is to capture consensus cluster structures via a two-stage learning scheme. Specifically, first learn similarity graph matrices of multiple views and then fuse them into a unified superior graph matrix. Most current methods learn pairwise similarities between data points for each view independently, which is widely used in single view. However, the consensus information contained in multiple views are ignored, and the involved biases lead to an undesirable unified graph matrix. To this end, we propose a bipartite graph based multi-view clustering (BIGMC) approach. The consensus information can be represented by a small number of representative uniform anchor points for different views. A bipartite graph is constructed between data points and the anchor points. BIGMC constructs the bipartite graph matrices of all views and fuses them to produce a unified bipartite graph matrix. The unified bipartite graph matrix in turn improves the bipartite graph similarity matrix of each view and updates the anchor points. The final unified graph matrix forms the final clusters directly. In BIGMC, an adaptive weight is added for each view to avoid outlier views. A low-rank constraint is imposed on the Laplacian matrix of the unified matrix to construct a multi-component unified bipartite graph, where the component number corresponds to the required cluster number. The objective function is optimized in an alternating optimization fashion. Experimental results on synthetic and real-world data sets demonstrate its effectiveness and superiority compared with the state-of-the-art baselines.

This paper concerns interference estimation problem for device-to-device (D2D) communication networks. In the considered system, D2D users share common spectrum resources, such that the D2D links have interference with each other. To achieve effective interference management, it is necessary to have an accurate understanding of the interference relationship of the D2D devices, which is difficult since the locations and the number of mobile devices can vary over time. In this paper, we formulate an interference estimation problem for the D2D communication network where the D2D users change over time. Our objective is to get accurate estimations for the interference suffered by a generic D2D link and for the interference a D2D link introduces on other users. We propose a graph convolutional neural network (GCN) based estimation model which can estimate the interference of a D2D link based on the location information of the corresponding D2D pairs. Simulation results show the performance of our method.

This paper concerns the cooperative spectrum sensing (CSS) for cognitive radio (CR) networks, where the secondary users (SUs) collaborate to detect the presence of the primary users (PUs). With CSS, the information from different SUs is first fused, then, the detection of the PU signal is implemented based on the fused information. Most of the previous works focus on the design of a mapping function that calculates the probability of the existence of the PU signal based on the fused information. In this paper, we study the fusion process which combines the information from all SUs based on the local property of each SU. A graph-based multi-view learning framework for CSS (GMCSS) is designed to better fuse the information from different SUs. In the proposed framework, the information from each SU is considered as a view of the state of the target wireless channel and is fused with the information from other SUs through a graph-based learning process. Simulation results demonstrate the effectiveness of our method.

Partial domain adaptation (PDA) aims to transfer knowledge from a label-rich source domain to a label-scarce target domain based on an assumption that the source label space subsumes the target label space. The major challenge is to promote positive transfer in the shared label space and circumvent negative transfer caused by the large mismatch across different label spaces. In this article, we propose a dual alignment approach for PDA (DAPDA), including three components: 1) a feature extractor extracts source and target features by the Siamese network; 2) a reweighting network produces "hard" labels, class-level weights for source features and "soft" labels, instance-level weights for target features; 3) a dual alignment network aligns intra domain and interdomain distributions. Specifically, the intra domain alignment aims to minimize the intraclass variances to enhance the intraclass compactness in both domains, and interdomain alignment attempts to reduce the discrepancies across domains by domain-wise and class-wise adaptations. The negative transfer can be alleviated by down-weighting source features with nonshared labels. The positive transfer can be enhanced by upweighting source features with shared labels. The adaptation can be achieved by minimizing the discrepancies based on class-weighted source data with hard labels and instance-weighed target data with soft labels. The effectiveness of our method has been demonstrated by outperforming state-of-the-art PDA methods on several benchmark datasets.

The existing unsupervised domain adaptation (UDA) methods require not only labeled source samples but also a large number of unlabeled target samples for domain adaptation. Collecting these target samples is generally time-consuming, which hinders the rapid deployment of these UDA methods in new domains. Besides, most of these UDA methods are developed for image classification. In this paper, we address a new problem called one-shot unsupervised domain adaptation for object detection, where only one unlabeled target sample is available. To the best of our knowledge, this is the first time this problem is investigated. To solve this problem, a one-shot feature alignment (OSFA) algorithm is proposed to align the low-level features of the source domain and the target domain. Specifically, the domain shift is reduced by aligning the average activation of the feature maps in the lower layer of CNN. The proposed OSFA is evaluated under two scenarios: adapting from clear weather to foggy weather; adapting from synthetic images to real-world images. Experimental results show that the proposed OSFA can significantly improve the object detection performance in target domain compared to the baseline model without domain adaptation.

In data mining, large differences between multi-class distributions regarded as class imbalance issues have been known to hinder the classification performance. Unfortunately, existing sampling methods have shown their deficiencies such as causing the problems of over-generation and over-lapping by oversampling techniques, or the excessive loss of significant information by undersampling techniques. This paper presents three proposed sampling approaches for imbalanced learning: the first one is the entropy-based oversampling (EOS) approach; the second one is the entropy-based undersampling (EUS) approach; the third one is the entropy-based hybrid sampling (EHS) approach combined by both oversampling and undersampling approaches. These three approaches are based on a new class imbalance metric, termed entropy-based imbalance degree (EID), considering the differences of information contents between classes instead of traditional imbalance-ratio. Specifically, to balance a data set after evaluating the information influence degree of each instance, EOS generates new instances around difficult-to-learn instances and only remains the informative ones. EUS removes easy-to-learn instances. While EHS can do both simultaneously. Finally, we use all the generated and remaining instances to train several classifiers. Extensive experiments over synthetic and real-world data sets demonstrate the effectiveness of our approaches.

We propose a novel cooperative spectrum sensing (CSS) framework for cognitive radio networks based on imbalanced learning techniques, which aims to resolve the skewed category distribution problems of signal data. For a radio channel shared by primary users (PUs) and secondary users (SUs), the signal data composed of energy vectors, in which each energy level is estimated by SU, can be used to detect the channel availability via a classifier. However, due to the nature of this application, the existing category-imbalance problem hinders the detection performance since the trained classifier has a better effect on the dominated category. To enhance the performance, sampling (e.g., oversampling, under-sampling, and combination) algorithms are employed to balance the training data set based on the imbalance degree metric of imbalance-ratio. The balanced training set then can be used to train classifiers with initial parameters, and the validation set can be utilized to tune as well as evaluate the classifiers. In the testing phase, the actual desired performance on unseen signal data can be determined based on the testing set, i.e., whether the channel is available or not. The performance of each sampling algorithm is measured in terms of receiver operating characteristic (ROC) curve and area under the ROC curve (AUC). The simulation results demonstrate the effectiveness of our proposed framework compared to traditional CSS methods.

Adversarial domain adaptation has achieved some success in learning transferable feature representations and reducing distribution discrepancy between source and target domains. However, existing approaches mainly focus on alignment of global source and target distributions without considering complex structures in categories underlying different distributions, resulting in domain confusion and the mix of distinguishable structures. In this paper, we propose an adversarial domain adaptation via category transfer (ADACT) approach for unsupervised domain adaptation (UDA). ADACT first captures multi-category information through training source and target feature generators as well as a label predictor. Secondly, it uses multi-category domain critic networks to category-wisely estimate Wasserstein distances across domains. Then it learns category-invariant feature representations by finely-grained matching different data distributions with the estimated Wasserstein distances. The adaptation can be achieved by the standard back-propagation training approach with this two-step iteration. The effectiveness of ADACT is demonstrated since it outperforms several state-of-the-art UDA methods on common domain adaptation datasets.

Gear safety factors are conducive to assessing the reliability of vehicle transmission gears. Due to the insufficiency and high coupling of existing gear data, the prediction of gear safety factors has long been a challenging issue in vehicle transmission industry through learning meaningful representations of high-dimensional gear parameters. This paper presents a framework to find a high-quality solution that improves prediction accuracy of gear safety factors. In the framework, to cope with the insufficiency of gear data, a generative model based on generative adversarial networks is established and proven to generate acceptable data with Adam optimizer. Then, eigen-error principal component analysis is proposed to extract features of gear parameters by reconstructing error function with eigenvectors and eigenvalues. Finally, particle swarm optimization and back propagation are applied to predict safety factors with these extracted features. Experimental results on real-world gear data of vehicle transmissions have validated the effectiveness of our proposed framework.

Fault diagnosis of a thermal system under varying operating conditions is of great importance for the safe and reliable operation of a power plant involved in peak shaving. However, it is a difficult task due to the lack of sufficient labeled data under some operating conditions. In practical applications, the model built on the labeled data under one operating condition will be extended to such operating conditions. Data distribution discrepancy can be triggered by variation of operating conditions and may degenerate the performance of the model. Considering the fact that data distributions are different but related under different operating conditions, this paper proposes a hierarchical deep domain adaptation (HDDA) approach to transfer a classifier trained on labeled data under one loading condition to identify faults with unlabeled data under another loading condition. In HDDA, a hierarchical structure is developed to reveal the effective information for final diagnosis by layerwisely capturing representative features. HDDA learns domain-invariant and discriminative features with the hierarchical structure by reducing distribution discrepancy and preserving discriminative information hidden in raw process data. For practical applications, the Taguchi method is used to obtain the optimized model parameters. Experimental results and comprehensive comparison analysis demonstrate its superiority.

A large number of datasets in various applications are imbalanced in which majority samples dominate minority samples. The skewed distribution poses a difficulty for existing learning approaches. Oversampling techniques address this concern by replicating original samples or adding new synthetic samples of minority class. Even with success, they raise the problems of over-generation and overlapping. In this paper, we propose an entropy difference-based oversampling approach (EDOS) for imbalanced learning using a novel metric, termed entropy difference (ED). First, given a dataset, EDOS measures the imbalance degree between the majority and the minority with ED. Second, EDOS creates synthetic minority samples. For each synthetic sample, EDOS evaluates its retention capability and remains the informative sample. Third, original and qualified synthetic samples are combined to train the classifiers. In the experiments, we demonstrate the effectiveness of the proposed EDOS method on several UCI datasets.

Clustering of data with high dimension and variable densities poses a remarkable challenge to the traditional density-based clustering methods. Recently, entropy, a numerical measure of the uncertainty of information, can be used to measure the border degree of samples in data space and also select significant features in feature set. It was used in our new framework based on the sparsity-density entropy (SDE) to cluster the data with high dimension and variable densities. First, SDE conducts high-quality sampling for multidimensional data and selects the representative features using sparsity score entropy (SSE). Second, the clustering results and noises are obtained adopting a new density-variable clustering method called density entropy (DE). DE automatically determines the border set based on the global minimum of border degrees and then adaptively performs cluster analysis for each local cluster based on the local minimum of border degrees. The effectiveness and efficiency of the proposed SDE framework are validated on synthetic and real data sets in comparison with several clustering algorithms. The results showed that the proposed SDE framework concurrently detected the noises and processed the data with high dimension and various densities.