In Zero-Shot Learning (ZSL), embedding-based methods learn a visual–semantic mapping that leverages the attribute knowledge of seen classes to predict the attributes of unseen classes, enabling knowledge transfer from seen to unseen classes. However, distributional discrepancies between seen and unseen classes introduce an inherent domain shift, and inter-class variations cause the same attribute to be expressed differently across categories. As a result, models trained on seen classes often struggle to accurately recognize attributes in unseen classes, limiting their generalization ability. To address these challenges, we propose DAPZSL, a dual adversarial perturbation framework that enhances the robustness of visual–semantic mappings through Feature-Level Adversarial Perturbation (FAP) and improves the model’s generalization ability via Weight-Level Adversarial Perturbation (WAP). Specifically, FAP generates semantically perturbed adversarial samples at the feature level, and incorporating these samples during training encourages the model to learn more robust visual–semantic mappings that are resilient to semantic variations, which improves attribute recognition on unseen classes. Meanwhile, WAP introduces adversarial perturbations into the model’s weight space, promoting a flatter loss landscape that alleviates overfitting to seen classes and enhances generalization. Extensive experiments on multiple benchmark datasets—including AWA2, SUN, and CUB—demonstrate that DAPZSL significantly outperforms existing ZSL models.

In distributed machine learning scenarios, the difference in data distribution among different nodes is a key issue that cannot be ignored. However, existing methods make it difficult to autonomously adjust model parameters for dynamically changing data distributions, leading to inflexible global decision boundaries with insufficient local adaptation. To address this problem, we propose a distributed hybrid support vector machine (SVM) based on the adaptive ensemble selection of regularization parameters, DHS-AE. The model utilizes the data structure information to cut the data space and thus identify data distribution characteristics. The SVM, integrated with regularization parameters that are adaptively determined within specific ranges, is utilized in the local subspace to enable real-time adjustment of decision boundaries in response to distribution changes, thereby further reducing the computational overhead. The generalization bound of DHS-AE is theoretically established using covering numbers, and the fast convergence speed and consistency are derived. In practical applications, we verify the excellent performance of the DHS-AE using a large number of real datasets.

Fuzzy logic-based graph neural networks (FL-GNN) have recently garnered growing attention in node classification, which aims to enhance the ability of GNN in modeling uncertain relationships between nodes. However, existing FL-GNN typically assume that nodes in the source domain (training set) and target domain (test set) follow the identical data distribution and class sets. Real-world scenarios often exhibit significant distribution shifts and target domain even contains classes that were not present in the source domain, termed open-set cross-domain node classification (OSCD-NC), which seriously damages their superior performance. Thus, how to leverage the strong uncertain knowledge representation capacity of FL-GNN to learn a well-defined boundary between seen and unseen classes for improving OSCD-NC performance remains an open and underexplored research problem. In this paper, we propose an effective domain-adaptive fuzzy graph diffusion network (DFGDN) for OSCD-NC. Specifically, with the help of a fuzzy adjacency matrix, fuzzy graph diffusion networks are proposed to generate robust fuzzy node representations by adaptively enhancing feature collaboration between low-pass and high-pass graph filters. Then, a peer ($M$+1)-class classifier is introduced to learn a rough class boundary by measuring their class prediction probability difference for target domain. After that, the ($M$+1)-means clustering and decoder modules are simultaneously designed to discover more supervision guidance from target domain for learned class boundary optimization. Finally, we jointly optimize the above modules in an adversarial manner via classification loss, classifier discrepancy loss and mean squared error loss, which further improves the accuracy of the learned class boundary by pulling seen nodes from the source domain and target domain closer, and pushing unseen nodes away. Extensive experiments on three cross-domain data pairs and various openness rates demonstrate the effectiveness of the proposed DFGDN framework.

Distributed machine learning provides an efficient solution for large-scale data processing through parallel computing. However, current distributed learning relies on global or local paradigms and cannot adaptively adjust decision boundaries in complex data environments. To address this problem, we propose a Distributed Hybrid Learning algorithm based on Fisher Linear Discriminant (DHL-FLD). Specifically, DHL-FLD consists of a global pre-learning phase and a subspace local learning phase. On the one hand, the global pre-learning phase is designed to divide the data space, which can obtain the data structure information. On the other hand, the local learning phase dynamically adjusts and optimizes the decision boundaries, guided by the structural information and distributional properties of the data. Theoretically, we establish the generalization bound of DHL-FLD using the integral operator technique and verify the scalability and robustness of DHL-FLD. The effectiveness of DHL-FLD is demonstrated through extensive experiments on real datasets.

With the fast development of AI-related techniques, the applications of trajectory prediction are no longer limited to easier scenes and trajectories. More and more trajectories with different forms, such as coordinates, bounding boxes, and even high-dimensional human skeletons, need to be analyzed and forecasted. Among these heterogeneous trajectories, interactions between different elements within a frame of trajectory, which we call "Dimension-wise Interactions", would be more complex and challenging. However, most previous approaches focus mainly on a specific form of trajectories, and potential dimension-wise interactions are less concerned. In this work, we expand the trajectory prediction task by introducing the trajectory dimensionality $M$M, thus extending its application scenarios to heterogeneous trajectories. We first introduce the Haar transform as an alternative to the Fourier transform to better capture the time-frequency properties of each trajectory-dimension. Then, we adopt the bilinear structure to model and fuse two factors simultaneously, including the time-frequency response and the dimension-wise interaction, to forecast heterogeneous trajectories via trajectory spectrums hierarchically in a generic way. Experiments show that the proposed model outperforms most state-of-the-art methods on ETH-UCY, SDD, nuScenes, and Human3.6 M with heterogeneous trajectories, including 2D coordinates, 2D/3D bounding boxes, and 3D human skeletons.

Unsupervised graph-structure learning (GSL) which aims to learn an effective graph structure applied to arbitrary downstream tasks by data itself without any labels' guidance, has recently received increasing attention in various real applications. Although several existing unsupervised GSL has achieved superior performance in different graph analytical tasks, how to utilize the popular graph masked autoencoder to sufficiently acquire effective supervision information from the data itself for improving the effectiveness of learned graph structure has been not effectively explored so far. To tackle the above issue, we present a multilevel contrastive graph masked autoencoder (MCGMAE) for unsupervised GSL. Specifically, we first introduce a graph masked autoencoder with the dual feature masking strategy to reconstruct the same input graph-structured data under the original structure generated by the data itself and learned graph-structure scenarios, respectively. And then, the inter- and intra-class contrastive loss is introduced to maximize the mutual information in feature and graph-structure reconstruction levels simultaneously. More importantly, the above inter- and intra-class contrastive loss is also applied to the graph encoder module for further strengthening their agreement at the feature-encoder level. In comparison to the existing unsupervised GSL, our proposed MCGMAE can effectively improve the training robustness of the unsupervised GSL via different-level supervision information from the data itself. Extensive experiments on three graph analytical tasks and eight datasets validate the effectiveness of the proposed MCGMAE.

Filter pruning has gained widespread adoption for the purpose of compressing and speeding up convolutional neural networks (CNNs). However, the existing approaches are still far from practical applications due to biased filter selection and heavy computation cost. This article introduces a new filter pruning method that selects filters in an interpretable, multiperspective, and lightweight manner. Specifically, we evaluate the contributions of filters from both individual and overall perspectives. For the amount of information contained in each filter, a new metric called information capacity is proposed. Inspired by the information theory, we utilize the interpretable entropy to measure the information capacity and develop a feature-guided approximation process. For correlations among filters, another metric called information independence is designed. Since the aforementioned metrics are evaluated in a simple but effective way, we can identify and prune the least important filters with less computation cost. We conduct comprehensive experiments on benchmark datasets employing various widely used CNN architectures to evaluate the performance of our method. For instance, on ILSVRC-2012, our method outperforms state-of-the-art methods by reducing floating-point operations (FLOPs) by 77.4% and parameters by 69.3% for ResNet-50 with only a minor decrease in an accuracy of 2.64%.

Sparse additive machines (SAMs) have shown competitive performance on variable selection and classification in high-dimensional data due to their representation flexibility and interpretability. However, the existing methods often employ the unbounded or nonsmooth functions as the surrogates of 0-1 classification loss, which may encounter the degraded performance for data with outliers. To alleviate this problem, we propose a robust classification method, named SAM with the correntropy-induced loss (CSAM), by integrating the correntropy-induced loss (C-loss), the data-dependent hypothesis space, and the weighted -norm regularizer ( ) into additive machines. In theory, the generalization error bound is estimated via a novel error decomposition and the concentration estimation techniques, which shows that the convergence rate can be achieved under proper parameter conditions. In addition, the theoretical guarantee on variable selection consistency is analyzed. Experimental evaluations on both synthetic and real-world datasets consistently validate the effectiveness and robustness of the proposed approach.

Recently emerged label noise-resistant graph representation learning (LNR-GRL) has received increasing attention, which aims to enhance the generalization of graph neural networks (GNNs) in semi-supervised node classification with noisy and limited labels. Most of the existing LNR-GRL tend to introduce more complex sample selection strategies developed in nongraph areas to distinguish more noisy nodes to alleviate their misguidance. However, these proposed methods neglect the importance of inaccurate graph structure relationships rectification, and information collaboration between inaccurate graph structure relationships and noisy node label rectification in improving the quality of noisy node identification and its rectified node labels. To solve the above-mentioned issues, we propose a novel multiplex experts governance collaboration (MEGC) framework for LNR-GRL. Specifically, an unsupervised graph structure governance expert is first designed to rectify inaccurate graph structure relationships. Based on the rectified graph structure, a simple label noise governance expert is proposed to accurately identify noisy node labels and further improve the quality of noisy nodes’ rectified labels and unlabeled nodes’ pseudo-labels. Finally, the above-proposed governance experts can be effectively combined with GNNs to jointly guide their training via the introduced cross-view graph contrastive loss and cross-entropy loss, which can maximally limit the effect of noisy node labels and discover more effective supervision guidance from data itself for GNNs optimization. Extensive experiments on three benchmarks, two label noise types, four noise rates, and four training label rates demonstrate the superiority of the proposed method in comparison to the existing LNR-GRL methods.

Few-shot graph anomaly detection (GAD) has recently garnered increasing attention, which aims to discern anomalous patterns among abundant unlabeled test nodes under the guidance of a limited number of labeled training nodes. Existing few-shot GAD approaches typically adopt meta-training methods trained on richly labeled auxiliary networks to facilitate rapid adaptation to target networks that possess sparse labels. However, these proposed methods often assume that the auxiliary and target networks exist in the same data distributions-an assumption rarely holds in practical settings. This paper explores a more prevalent and complex scenario of cross-domain few-shot GAD, where the goal is to identify anomalies within sparsely labeled target graphs using auxiliary graphs from a related, yet distinct domain. The challenge here is nontrivial owing to inherent data distribution discrepancies between the source and target domains, compounded by the uncertainties of sparse labeling in the target domain. In this paper, we propose a simple and effective framework, termed CDFS-GAD, specifically designed to tackle the aforementioned challenges. CDFS-GAD first introduces a domain-adaptive graph contrastive learning module, which is aimed at enhancing cross-domain feature alignment. Then, a prompt tuning module is further designed to extract domain-specific features tailored to each domain. Moreover, a domain-adaptive hypersphere classification loss is proposed to enhance the discrimination between normal and anomalous instances under minimal supervision, utilizing domain-sensitive norms. Lastly, a self-training strategy is introduced to further refine the predicted scores, enhancing its reliability in few-shot settings. Extensive experiments on twelve real-world cross-domain data pairs demonstrate the effectiveness of the proposed CDFS-GAD framework in comparison to various existing GAD methods including unsupervised, semi-supervised, few-shot and cross-domain GAD methods.

Generalized additive models (GAMs) have been successfully applied to high dimensional data. However, most existing methods cannot capture the high level feature patterns from complex data. To alleviate this problem, we propose a new sparse additive model, named generalized sparse additive model with unknown link function (GSAMUL), in which the component functions are estimated by B-spline basis and the unknown link function is estimated by a multi-layer perceptron (MLP) network. Furthermore,$\mathscr{l}_{2.1}$-norm regularizer is used for variable selection. The proposed GSAMUL can realize both variable selection and hidden interaction. We integrate this estimation into a bilevel optimization problem, where the data is split into training set and validation set. In theory, we provide the guarantees about the convergence of the approximate procedure. In applications, experimental evaluations on both synthetic and real world data sets consistently validate the effectiveness of GSAMUL.

Weakly-supervised fine-grained visual categorization (FGVC) aims to achieve subclass classification within the same large class using only label information. Compared to general images, fine-grained images have similar appearances and features, and are often affected by disturbances such as viewpoint, lighting, and occlusion during data collection, resulting in significant intra-class variance and small inter-class variance. To achieve FGVC, carefully designed models are often needed to explore the locally discriminative regions of the image. This paper revisits high-quality FGVC publications based on deep learning and analyzes from two new perspective: fine-grained image data and backbone. We address two ignored but interesting problems in FGVC. First, we argue that the reasons for exacerbating intra-class variance are not the same in data of animal, plant, and commodity types, and it is necessary to consider the effects of posture, covariate shift, and structural changes. Additionally, the “soft boundary” between subclasses intensifies the difficulty of classification. Second, we highlight that convolutional networks and self-attention networks have different receptive fields and shape biases, leading to performance differences when processing different types of fine-grained data. Overall, our analysis provides new insights into recent advances, challenges, and future directions for FGVC based on deep learning, which can help researchers develop more effective models for FGVC.

Few-shot class-incremental learning has recently received significant research focus in remote sensing scene classification (FSCIL-RSSC). The success of FSCIL-RSSC relies on the robustness of the feature backbone and classifiers. Existing works focus on improving classifier adaptation, but little attention is paid to the importance of backbone robustness on the recognition ability of new class samples’ embeddings. Due to the large distribution shift between old and new classes, FSCIL-RSSC using high-layer (single-scale) features may not adapt flawlessly to new categories. To solve the issue, we put forward a gradient guided multiscale feature collaboration network (G-MFCN) for FSCIL-RSSC. Specifically, we introduce a parallel hierarchy strategy to simultaneously capture the multifeature discriminative information of the same sample. Then, a gradient guide block is designed to automatically pick out the optimal values of different convolution blocks for multifeature fusion. Finally, the classical feature pyramid network is introduced for multiscale fusion to obtain more obvious discriminative features of RSSC. More importantly, our proposed G-MFCN is a simple and adaptable module, which can combine any existing FSCIL frameworks to further improve the optimized classifiers’ effectiveness for the FSCIL-RSSC scenario. Extensive experiments on four benchmarks demonstrate that the proposed G-MFCN achieves significant improvements in comparison to existing FSCIL-RSSC methods.

Class-incremental (CI) learning has recently received extensive research interest in remote sensing scene classification (CI-RSSC). The existing CI-RSSC methods’ superior performance seriously relies on old (base classes) and new classes (incremental classes) sampled independently from an identical distribution (dataset). In real-world RSSC scenarios, there exist significant distribution shifts between old and new classes, leading to the existing CI-RSSC methods being unable to adjust flawlessly to these new classes. In this article, we propose a novel cross-domain (CD) CI-RSSC framework to solve the above-mentioned problems, termed CDCI-RSSC. Specifically, a modular sharing-based dynamic extension module is first designed, which only updates specialized modules to extract new class feature embeddings for reducing memory footprint. Then, an effective dynamic alignment guided domain adaptive module (DAM) is further proposed to calculate the dynamic weights of each sample in various fields, which can minimize distribution shifts between source and target domains. Finally, a foreground enhancement module (FEM) is introduced to alleviate the issue of complex background interference in RSSC by increasing the weight of critical regions. Compared with the existing CI-RSSC and CD-RSSC, our proposed CDCI-RSSC framework surmounts the challenge of handling the distribution shifts between source (base session) and target domains (incremental session) while alleviating the limitations of continuous learning of new classes. Extensive experiments on three CDCI scenarios show that the CDCI-RSSC model achieves significant performance improvements in comparison to existing CI-RSSC and CD-RSSC methods.

Zero-shot learning (ZSL) tackles the unseen class recognition problem by transferring semantic knowledge from seen classes to unseen ones. Typically, to guarantee desirable knowledge transfer, a direct embedding is adopted for associating the visual and semantic domains in ZSL. However, most existing ZSL methods focus on learning the embedding from implicit global features or image regions to the semantic space. Thus, they fail to: 1) exploit the appearance relationship priors between various local regions in a single image, which corresponds to the semantic information and 2) learn cooperative global and local features jointly for discriminative feature representations. In this article, we propose the novel graph navigated dual attention network (GNDAN) for ZSL to address these drawbacks. GNDAN employs a region-guided attention network (RAN) and a region-guided graph attention network (RGAT) to jointly learn a discriminative local embedding and incorporate global context for exploiting explicit global embeddings under the guidance of a graph. Specifically, RAN uses soft spatial attention to discover discriminative regions for generating local embeddings. Meanwhile, RGAT employs an attribute-based attention to obtain attribute-based region features, where each attribute focuses on the most relevant image regions. Motivated by the graph neural network (GNN), which is beneficial for structural relationship representations, RGAT further leverages a graph attention network to exploit the relationships between the attribute-based region features for explicit global embedding representations. Based on the self-calibration mechanism, the joint visual embedding learned is matched with the semantic embedding to form the final prediction. Extensive experiments on three benchmark datasets demonstrate that the proposed GNDAN achieves superior performances to the state-of-the-art methods. Our code and trained models are available at https://github.com/shiming-chen/GNDAN.

Despite the great success of the existing work in fine-grained visual categorization (FGVC), there are still several unsolved challenges, e.g., poor interpretation and vagueness contribution. To circumvent this drawback, motivated by the hypersphere embedding method, we propose a discriminative suprasphere embedding (DSE) framework, which can provide intuitive geometric interpretation and effectively extract discriminative features. Specifically, DSE consists of three modules. The first module is a suprasphere embedding (SE) block, which learns discriminative information by emphasizing weight and phase. The second module is a phase activation map (PAM) used to analyze the contribution of local descriptors to the suprasphere feature representation, which uniformly highlights the object region and exhibits remarkable object localization capability. The last module is a class contribution map (CCM), which quantitatively analyzes the network classification decision and provides insight into the domain knowledge about classified objects. Comprehensive experiments on three benchmark datasets demonstrate the effectiveness of our proposed method in comparison with state-of-the-art methods.

Graph neural networks (GNNs) with missing node features have recently received increasing interest. Such missing node features seriously hurt the performance of the existing GNNs. Some recent methods have been proposed to reconstruct the missing node features by the information propagation among nodes with known and unknown attributes. Although these methods have achieved superior performance, how to exactly exploit the complex data correlations among nodes to reconstruct missing node features is still a great challenge. To solve the above problem, we propose a self-supervised guided hypergraph feature propagation (SGHFP). Specifically, the feature hypergraph is first generated according to the node features with missing information. And then, the reconstructed node features produced by the previous iteration are fed to a two-layer GNNs to construct a pseudo-label hypergraph. Before each iteration, the constructed feature hypergraph and pseudo-label hypergraph are fused effectively, which can better preserve the higher-order data correlations among nodes. After then, we apply the fused hypergraph to the feature propagation for reconstructing missing features. Finally, the reconstructed node features by multi-iteration optimization are applied to the downstream semi-supervised classification task. Extensive experiments demonstrate that the proposed SGHFP outperforms the existing semi-supervised classification with missing node feature methods.

In recent years, graph neural networks (GNN) have achieved significant developments in a variety of graph analytical tasks. Nevertheless, GNN’s superior performance will suffer from serious damage when the collected node features or structure relationships are partially missing owning to numerous unpredictable factors. Recently emerged graph completion learning (GCL) has received increasing attention, which aims to reconstruct the missing node features or structure relationships under the guidance of a specifically supervised task. Although these proposed GCL methods have made great success, they still exist the following problems: the reliance on labels, the bias of the reconstructed node features and structure relationships. Besides, the generalization ability of the existing GCL still faces a huge challenge when both collected node features and structure relationships are partially missing at the same time. To solve the above issues, we propose a more general GCL framework with the aid of self-supervised learning for improving the task performance of the existing GNN variants on graphs with features and structure missing, termed unsupervised GCL (UGCL). Specifically, to avoid the mismatch between missing node features and structure during the message-passing process of GNN, we separate the feature reconstruction and structure reconstruction and design its personalized model in turn. Then, a dual contrastive loss on the structure level and feature level is introduced to maximize the mutual information of node representations from feature reconstructing and structure reconstructing paths for providing more supervision signals. Finally, the reconstructed node features and structure can be applied to the downstream node classification task. Extensive experiments on eight datasets demonstrate the effectiveness of our proposed method.

Few-shot class-incremental learning (FSCIL) has recently attracted extensive attention in various areas. Existing FSCIL methods highly depend on the robustness of the feature backbone pre-trained on base classes. In recent years, different Transformer variants have obtained significant processes in the feature representation learning of massive fields. Nevertheless, the progress of the Transformer in FSCIL scenarios has not achieved the potential promised in other fields so far. In this paper, we develop a semantic-visual guided Transformer (SV-T) to enhance the feature extracting capacity of the pre-trained feature backbone on incremental classes. Specifically, we first utilize the visual (image) labels provided by the base classes to supervise the optimization of the Transformer. And then, a text encoder is introduced to automatically generate the corresponding semantic (text) labels for each image from the base classes. Finally, the constructed semantic labels are further applied to the Transformer for guiding its hyperparameters updating. Our SV-T can take full advantage of more supervision information from base classes and further enhance the training robustness of the feature backbone. More importantly, our SV-T is an independent method, which can directly apply to the existing FSCIL architectures for acquiring embeddings of various incremental classes. Extensive experiments on three benchmarks, two FSCIL architectures, and two Transformer variants show that our proposed SV-T obtains a significant improvement in comparison to the existing state-of-the-art FSCIL methods.

Trajectory prediction aims to forecast agents’ possible future locations considering their observations along with the video context. It is strongly needed by many autonomous platforms like tracking, detection, robot navigation, and self-driving cars. Whether it is agents’ internal personality factors, interactive behaviors with the neighborhood, or the influence of surroundings, they all impact agents’ future planning. However, many previous methods model and predict agents’ behaviors with the same strategy or feature distribution, making them challenging to make predictions with sufficient style differences. This paper proposes the Multi-Style Network (MSN), which utilizes style proposal and stylized prediction using two sub-networks, to provide multi-style predictions in a novel categorical way adaptively. The proposed network contains a series of style channels, and each channel is bound to a unique and specific behavior style. We use agents’ end-point plannings and their interaction context as the basis for the behavior classification, so as to adaptively learn multiple diverse behavior styles through these channels. Then, we assume that the target agents may plan their future behaviors according to each of these categorized styles, thus utilizing different style channels to make predictions with significant style differences in parallel. Experiments show that the proposed MSN outperforms current state-of-the-art methods up to 10% quantitatively on two widely used datasets, and presents better multi-style characteristics qualitatively.

Recent weakly supervised semantic segmentation methods generate pseudolabels to recover the lost position information in weak labels for training the segmentation network. Unfortunately, those pseudolabels often contain mislabeled regions and inaccurate boundaries due to the incomplete recovery of position information. It turns out that the result of semantic segmentation becomes determinate to a certain degree. In this article, we decompose the position information into two components: high-level semantic information and low-level physical information, and develop a componentwise approach to recover each component independently. Specifically, we propose a simple yet effective pseudolabels updating mechanism to iteratively correct mislabeled regions inside objects to precisely refine high-level semantic information. To reconstruct low-level physical information, we utilize a customized superpixel-based random walk mechanism to trim the boundaries. Finally, we design a novel network architecture, namely, a dual-feedback network (DFN), to integrate the two mechanisms into a unified model. Experiments on benchmark datasets show that DFN outperforms the existing state-of-the-art methods in terms of intersection-over-union (mIoU).

Dynamic texture (DT) exhibits statistical stationarity in the spatial domain and stochastic repetitiveness in the temporal dimension, indicating that different frames of DT possess a high similarity correlation that is critical prior knowledge. However, existing methods cannot effectively learn a synthesis model for high-dimensional DT from a small number of training samples. In this article, we propose a novel DT synthesis method, which makes full use of similarity as prior knowledge to address this issue. Our method is based on the proposed kernel similarity embedding, which can not only mitigate the high dimensionality and small sample issues, but also has the advantage of modeling nonlinear feature relationships. Specifically, we first put forward two hypotheses that are essential for the DT model to generate new frames using similarity correlations. Then, we integrate kernel learning and the extreme learning machine into a unified synthesis model to learn kernel similarity embeddings for representing DTs. Extensive experiments on DT videos collected from the Internet and two benchmark datasets, i.e., Gatech Graphcut Textures and Dyntex, demonstrate that the learned kernel similarity embeddings can provide discriminative representations for DTs. Further, our method can preserve the long-term temporal continuity of the synthesized DT sequences with excellent sustainability and generalization. Meanwhile, it effectively generates realistic DT videos with higher speed and lower computation than the current state-of-the-art methods. The code and more synthesis videos are available at our project pagehttps://shiming-chen.github.io/Similarity-page/Similarit.html.

Zero-shot learning (ZSL) aims to recognize novel classes by transferring semantic knowledge from seen classes to unseen ones. Semantic knowledge is learned from attribute descriptions shared between different classes, which are strong prior for localization of object attribute for representing discriminative region features enabling significant visual-semantic interaction. Although few attention-based models have attempted to learn such region features in a single image, the transferability and discriminative attribute localization of visual features are typically neglected. In this paper, we propose an attribute-guided Transformer network to learn the attribute localization for discriminative visual-semantic embedding representations in ZSL, termed TransZero. Specifically, TransZero takes a feature augmentation encoder to alleviate the cross-dataset bias between ImageNet and ZSL benchmarks and improve the transferability of visual features by reducing the entangled relative geometry relationships among region features. To learn locality-augmented visual features, TransZero employs a visual-semantic decoder to localize the most relevant image regions to each attributes from a given image under the guidance of attribute semantic information. Then, the locality-augmented visual features and semantic vectors are used for conducting effective visual-semantic interaction in a visual-semantic embedding network. Extensive experiments show that TransZero achieves a new state-of-the-art on three ZSL benchmarks. The codes are available at: https://github.com/shiming-chen/TransZero.

The key challenge of zero-shot learning (ZSL) is how to infer the latent semantic knowledge between visual and attribute features on seen classes, and thus achieving a desirable knowledge transfer to unseen classes. Prior works either simply align the global features of an image with its associated class semantic vector or utilize unidirectional attention to learn the limited latent semantic representations, which could not effectively discover the intrinsic semantic knowledge (e.g., attribute semantics) between visual and attribute features. To solve the above dilemma, we propose a Mutually Semantic Distillation Network (MSDN), which progressively distills the intrinsic semantic representations between visual and attribute features for ZSL. MSDN incorporates an attribute→visual attention sub-net that learns attribute-based visual features, and a visual→attribute attention sub-net that learns visual-based attribute features. By further introducing a semantic distillation loss, the two mutual attention sub-nets are capable of learning collaboratively and teaching each other throughout the training process. The proposed MSDN yields significant improvements over the strong baselines, leading to new state-of-the-art performances on three popular challenging benchmarks. Our codes have been available at: https://github.com/shiming-chen/MSDN.

Generative methods have been successfully applied in zero-shot learning (ZSL) by learning an implicit mapping to alleviate the visual-semantic domain gaps and synthesizing unseen samples to handle the data imbalance between seen and unseen classes. However, existing generative methods simply use visual features extracted by the pre-trained CNN backbone. These visual features lack attribute-level semantic information. Consequently, seen classes are indistinguishable, and the knowledge transfer from seen to unseen classes is limited. To tackle this issue, we propose a novel Semantic Compression Embedding Guided Generation (SC-EGG) model, which cascades a semantic compression embedding network (SCEN) and an embedding guided generative network (EGGN). The SCEN extracts a group of attribute-level local features for each sample and further compresses them into the new low-dimension visual feature. Thus, a dense-semantic visual space is obtained. The EGGN learns a mapping from the class-level semantic space to the dense-semantic visual space, thus improving the discriminability of the synthesized dense-semantic unseen visual features. Extensive experiments on three benchmark datasets, i.e., CUB, SUN and AWA2, demonstrate the significant performance gains of SC-EGG over current state-of-the-art methods and its baselines.

Cross-modal hashing that leverages hash functions to project high-dimensional data from different modalities into the compact common hamming space, has shown immeasurable potential in cross-modal retrieval. To ease labor costs, unsupervised cross-modal hashing methods are proposed. However, existing unsupervised methods still suffer from two factors in the optimization of hash functions: 1) similarity guidance, they barely give a clear definition of whether is similar or not between data points, leading to the residual of the redundant information; 2) optimization strategy, they ignore the fact that the similarity learning abilities of different hash functions are different, which makes the hash function of one modality weaker than the hash function of the other modality. To alleviate such limitations, this paper proposes an unsupervised cross-modal hashing method to train hash functions with discriminative similarity guidance and adaptively-enhanced optimization strategy, termed Deep Adaptively-Enhanced Hashing (DAEH). Specifically, to estimate the similarity relations with discriminability, Information Mixed Similarity Estimation (IMSE) is designed by integrating information from distance distributions and the similarity ratio. Moreover, Adaptive Teacher Guided Enhancement (ATGE) optimization strategy is also designed, which employs information theory to discover the weaker hash function and utilizes an extra teacher network to enhance it. Extensive experiments on three benchmark datasets demonstrate the superiority of the proposed DAEH against the state-of-the-arts.

Generalized zero-shot learning (GZSL) has achieved significant progress, with many efforts dedicated to over-coming the problems of visual-semantic domain gap and seen-unseen bias. However, most existing methods directly use feature extraction models trained on ImageNet alone, ignoring the cross-dataset bias between ImageNet and GZSL benchmarks. Such a bias inevitably results in poor-quality visual features for GZSL tasks, which potentially limits the recognition performance on both seen and unseen classes. In this paper, we propose a simple yet effective GZSL method, termed feature refinement for generalized zero-shot learning (FREE), to tackle the above problem. FREE employs a feature refinement (FR) module that in-corporates semantic→visual mapping into a unified generative model to refine the visual features of seen and unseen class samples. Furthermore, we propose a self-adaptive margin center loss (SAMC-loss) that cooperates with a semantic cycle-consistency loss to guide FR to learn class- and semantically-relevant representations, and concatenate the features in FR to extract the fully refined features. Extensive experiments on five benchmark datasets demonstrate the significant performance gain of FREE over its baseline and current state-of-the-art methods. The code is available at https://github.com/shiming-chen/FREE.

Zero-shot sketch-based image retrieval (ZS-SBIR), which aims to retrieve photos with sketches under the zero-shot scenario, has shown extraordinary talents in real-world applications. Most existing methods leverage language models to generate class-prototypes and use them to arrange the locations of all categories in the common space for photos and sketches. Although great progress has been made, few of them consider whether such pre-defined prototypes are necessary for ZS-SBIR, where locations of unseen class samples in the embedding space are actually determined by visual appearance and a visual embedding actually performs better. To this end, we propose a novel Norm-guided Adaptive Visual Embedding (NAVE) model, for adaptively building the common space based on visual similarity instead of language-based pre-defined prototypes. To further enhance the representation quality of unseen classes for both photo and sketch modality, modality norm discrepancy and noisy label regularizer are jointly employed to measure and repair the modality bias of the learned common embedding. Experiments on two challenging datasets demonstrate the superiority of our NAVE over state-of-the-art competitors.

Zero-shot learning (ZSL) tackles the unseen class recognition problem, transferring semantic knowledge from seen classes to unseen ones. Typically, to guarantee desirable knowledge transfer, a common (latent) space is adopted for associating the visual and semantic domains in ZSL. However, existing common space learning methods align the semantic and visual domains by merely mitigating distribution disagreement through one-step adaptation. This strategy is usually ineffective due to the heterogeneous nature of the feature representations in the two domains, which intrinsically contain both distribution and structure variations. To address this and advance ZSL, we propose a novel hierarchical semantic-visual adaptation (HSVA) framework. Specifically, HSVA aligns the semantic and visual domains by adopting a hierarchical two-step adaptation, i.e., structure adaptation and distribution adaptation. In the structure adaptation step, we take two task-specific encoders to encode the source data (visual domain) and the target data (semantic domain) into a structure-aligned common space. To this end, a supervised adversarial discrepancy (SAD) module is proposed to adversarially minimize the discrepancy between the predictions of two task-specific classifiers, thus making the visual and semantic feature manifolds more closely aligned. In the distribution adaptation step, we directly minimize the Wasserstein distance between the latent multivariate Gaussian distributions to align the visual and semantic distributions using a common encoder. Finally, the structure and distribution adaptation are derived in a unified framework under two partially-aligned variational autoencoders. Extensive experiments on four benchmark datasets demonstrate that HSVA achieves superior performance on both conventional and generalized ZSL. The code is available at \url{https://github.com/shiming-chen/HSVA}.

Considering that multiple features can improve the classification performance as they contain diversity information of images, a multiple-feature latent space learning-based method is proposed for hyperspectral image (HSI) classification in this letter. In the proposed method, a latent space that contains diversity information of multiple features and transformation matrices between the latent space and features are both learned. Moreover, spatial information is used for labeling unlabeled samples in the classification. Experimental results on the Indian Pines and University of Pavia data sets demonstrate the effectiveness of the proposed method.

Generative adversarial networks (GANs) have been a popular deep generative model for real-world applications. Despite many recent efforts on GANs that have been contributed, mode collapse and instability of GANs are still open problems caused by their adversarial optimization difficulties. In this article, motivated by the cooperative co-evolutionary algorithm, we propose a cooperative dual evolution-based GAN (CDE-GAN) to circumvent these drawbacks. In essence, CDE-GAN incorporates dual evolution with respect to the generator(s) and discriminators into a unified evolutionary adversarial framework to conduct effective adversarial multiobjective optimization. Thus, it exploits the complementary properties and injects dual mutation diversity into the training, to steadily diversify the estimated density in capturing multimodes and improve generative performance. Specifically, CDE-GAN decomposes the complex adversarial optimization problem into two subproblems (generation and discrimination), and each subproblem is solved with a separated subpopulation (E-GeneratorsandE-Discriminators), evolved by its own evolutionary algorithm. Additionally, we further propose aSoft Mechanismto balance the tradeoff between E-Generators and E-Discriminators to conduct steady training for CDE-GAN. Extensive experiments on one synthetic dataset and three real-world benchmark image datasets demonstrate that the proposed CDE-GAN achieves a competitive and superior performance in generating good quality and diverse samples over baselines. The code and more generated results are available at our project homepagehttps://shiming-chen.github.io/CDE-GAN-website/CDE-GAN.html.

Due to the high dimensionality of hyperspectral images (HSIs), more training samples are needed in general for better classification performance. However, surface materials cannot always provide sufficient training samples in practice. HSI classification with small size training samples is still a challenging problem. Multiview learning is a feasible way to improve the classification accuracy in the case of small training samples by combining information from different views. This article proposes a new spatial window-based multiview intact feature learning method (SWMIFL) for HSI classification. In the proposed SWMIFL, multiple features that reflect different information of the original image are extracted and spatial windows are imposed on training samples to select unlabeled samples. Then, multiview intact feature learning is performed to learn the intact feature of the training and unlabeled samples. Considering that neighboring samples are likely to belong to the same class, labels of spatial neighboring samples are determined by two factors including the labels of training samples that locate in the spatial window and the labels learned from the intact feature. Finally, unlabeled samples that have same labels under these two factors are treated as new training samples. Experimental results demonstrate that the proposed SWMIFL-based classification method outperforms several well-known HSI classification methods on three real-world data sets.

Low-rank Multiview Subspace Learning (LMvSL) has shown great potential in cross-view classification in recent years. Despite their empirical success, existing LMvSL-based methods are incapable of handling well view discrepancy and discriminancy simultaneously, which, thus, leads to performance degradation when there is a large discrepancy among multiview data. To circumvent this drawback, motivated by the block-diagonal representation learning, we propose structured low-rank matrix recovery (SLMR), a unique method of effectively removing view discrepancy and improving discriminancy through the recovery of the structured low-rank matrix. Furthermore, recent low-rank modeling provides a satisfactory solution to address the data contaminated by the predefined assumptions of noise distribution, such as Gaussian or Laplacian distribution. However, these models are not practical, since complicated noise in practice may violate those assumptions and the distribution is generally unknown in advance. To alleviate such a limitation, modal regression is elegantly incorporated into the framework of SLMR (termed MR-SLMR). Different from previous LMvSL-based methods, our MR-SLMR can handle any zero-mode noise variable that contains a wide range of noise, such as Gaussian noise, random noise, and outliers. The alternating direction method of multipliers (ADMM) framework and half-quadratic theory are used to optimize efficiently MR-SLMR. Experimental results on four public databases demonstrate the superiority of MR-SLMR and its robustness to complicated noise.

This article presents a new user-independent emotion classification method that classifies four distinct emotions using electroencephalograph (EEG) signals and the broad learning system (BLS). The public DEAP and MAHNOB-HCI databases are used. Just one EEG electrode channel is selected for the feature extraction process. Continuous wavelet transform (CWT) is then utilized to extract the proposed gray-scale image (GSI) feature which describes the EEG brain activation in both time and frequency domains. Finally, the new BLS is constructed for the emotion classification process, which successfully upgrades the efficiency of emotion classification based on EEG brain signals. The experiment results show that the proposed work produces a robust system with high accuracy of approximately 93.1% and training process time of approximately 0.7 s for the DEAP database, as well as, the high average accuracy of approximately 94.4% and training process time of approximately 0.6 s for MAHNOB-HCI database.

Cross-modal hashing intends to project data from two modalities into a common hamming space to perform cross-modal retrieval efficiently. Despite satisfactory performance achieved on real applications, existing methods are incapable of effectively preserving semantic structure to maintain inter-class relationship and improving discriminability to make intra-class samples aggregated simultaneously, which thus limits the higher retrieval performance. To handle this problem, we propose Equally-Guided Discriminative Hashing (EGDH), which jointly takes into consideration semantic structure and discriminability. Specifically, we discover the connection between semantic structure preserving and discriminative methods. Based on it, we directly encode multi-label annotations that act as high-level semantic features to build a common semantic structure preserving classifier. With the common classifier to guide the learning of different modal hash functions equally, hash codes of samples are intra-class aggregated and inter-class relationship preserving. Experimental results on two benchmark datasets demonstrate the superiority of EGDH compared with the state-of-the-arts.

We present an approach to extract the salient object automatically in videos. Given an unannotated video sequence, the proposed method first computes the visual saliency to identify object-like regions in each frame based on the proposed weighted multiple manifold ranking algorithm. We then compute motion cues to estimate the motion saliency and localization prior. Finally, adopting a new energy function, we estimate a superpixel-level object labeling across all frames, where 1) the data term depends on the visual saliency and localization prior, and 2) the smoothness term depends on the constraints in time and space. Compared to the existing counterparts, the proposed approach automatically segments the persistent foreground object meanwhile preserving the potential shape. Experiments show its promising results on the challenging benchmark videos in comparison with the existing counterparts.

Facial retouching has been increasingly applied in current social media and entertainment industries. In this study, the authors propose an efficient approach to automatically detect and retouch the facial flaws by using discriminative structure tensor. First, a non‐linear structure tensor associated with saliency model is exploited to discriminatively and automatically detect the significant facial flaws. Then, a Gaussian skin model is constructed in YCbCr space and the OSTU operation is simultaneously utilised to precisely mark the facial skin regions, in which the mouth, eyebrows and nostril parts are excluded. Subsequently, diverse structure tensor is employed to discriminatively adjust the inpainting priority and propose a structure tensor‐based inpainting algorithm to retouch the detected flaws. Without manual intervention, the extensive experiments have shown its effectiveness in marking the freckles, blemishes and moles in face images, and the retouching performance is visually pleasing in comparison with state‐of‐the‐art counterparts.

It is usually hard to predetermine the true number of segments in lip segmentation. This paper, therefore, presents a clustering-based approach to lip segmentation without knowing the true segment number. The objective function in the proposed approach is a variant of the partition entropy (PE) and features that the coincident cluster centroids in pattern space can be equivalently substituted by one centroid with the function value unchanged. It is shown that the minimum of the proposed objective function can be reached provided that: 1) the number of positions occupied by cluster centroids in pattern space is equal to the true number of clusters and 2) these positions are coincident with the optimal cluster centroids obtained under PE criterion. In implementation, we first randomly initialize the clusters provided that the number of clusters is greater than or equal to the ground truth. Then, an iterative algorithm is utilized to minimize the proposed objective function. For each iterative step, not only is the winner, i.e., the centroid with the maximum membership degree, updated to adapt to the corresponding input data, but also the other centroids are adjusted with a specific cooperation strength, so that they are each close to the winner. Subsequently, the initial overpartition will be gradually faded out with the redundant centroids superposed over the convergence of the algorithm. Based upon the proposed algorithm, we present a lip segmentation scheme. Empirical studies have shown its efficacy in comparison with the existing methods.

This paper presents a visual saliency detection approach, which is a hybrid of local feature-based saliency and global feature-based saliency (simply called local saliency and global saliency, respectively, for short). First, we propose an automatic selection of smoothing parameter scheme to make the foreground and background of an input image more homogeneous. Then, we partition the smoothed image into a set of regions and compute the local saliency by measuring the color and texture dissimilarity in the smoothed regions and the original regions, respectively. Furthermore, we utilize the global color distribution model embedded with color coherence, together with the multiple edge saliency, to yield the global saliency. Finally, we combine the local and global saliencies, and utilize the composition information to obtain the final saliency. Experimental results show the efficacy of the proposed method, featuring: 1) the enhanced accuracy of detecting visual salient region and appearance in comparison with the existing counterparts, 2) the robustness against the noise and the low-resolution problem of images, and 3) its applicability to multisaliency detection task.

This paper addresses the eye gaze tracking problem using a low cost and more convenient web camera in a desktop environment, as opposed to gaze tracking techniques requiring specific hardware, e.g, infrared high-resolution camera and infrared light sources, as well as a cumbersome calibration process. In the proposed method, we first track the human face in a real-time video sequence to extract the eye regions. Then, we combine intensity energy and edge strength to obtain the iris center and utilize the piecewise eye corner detector to detect the eye corner. We adopt a sinusoidal head model to simulate the 3-D head shape, and propose an adaptive weighted facial features embedded in the pose from the orthography and scaling with iterations algorithm, whereby the head pose can be estimated. Finally, the eye gaze tracking is accomplished by integration of the eye vector and the head movement information. Experiments are performed to estimate the eye movement and head pose on the BioID dataset and pose dataset, respectively. In addition, experiments for gaze tracking are performed in real-time video sequences under a desktop environment. The proposed method is not sensitive to the light conditions. Experimental results show that our method achieves an average accuracy of around 1.28° without head movement and 2.27° with minor movement of the head.

Outlier often degrades the classification and cluster accuracy. In this paper, we present an outlier detection approach based on local kernel regression for instance selection. It evaluates the reconstruction error of instances by their neighbors to identify the outliers. Experiments are performed both on the synthetic and real-life data sets to show the efficacy of the proposed approach in comparison with the existing counterparts.

The importance of high-fidelity enhancement in low quality fingerprint image cannot be overemphasized. Most of the existing fingerprint enhancement methods are contextual filter-based methods and they often suffer from two shortcomings: (1) there is block effect on the enhanced images; and (2) they blur or destroy ridge structures around singular points. In order to well preserve the ridge structures in singular regions and avoid block effect, we develop a new method for fingerprint enhancement combining nontensor product wavelet filter banks and anisotropic filter. We first decompose the fingerprint image using the nontensor product wavelet filter banks. Then we modify the approximation subimage using anisotropic filtering and adjust the high frequency coefficients of the three other subimages by applying the adaptive approach to reduce the noises according to the geometry feature of images. Finally, the inverse transform is applied to map the result and a final contrast enhancement is done subsequently. Experiments have been conducted on the fingerprint database FVC2004 in our study. The results demonstrate that the proposed approach is capable of overcoming block effect and enhancing low quality fingerprint while preserving the ridge structures around singular points.

The low-contrast and narrow blood vessels in retinal images are difficult to be extracted but useful in revealing certain systemic disease. Motivated by the goals of improving detection of such vessels, we propose the radial projection method to locate the vessel centerlines. Then the supervised classification is used for extracting the major structures of vessels. The final segmentation is obtained by the union of the two types of vessels after removal schemes. Our approach is tested on the STARE database, the results demonstrate that our algorithm can yield better segmentation.