Semi-supervised hyperspectral image (HSI) classification methods focus on exploring the spectral and spatial information of unlabeled samples. However, existing methods generally follow the closed-set setting, assuming that unlabeled samples do not contain novel classes, which is hard to hold in practical applications. This paper aims to study semi-supervised HSI classification in the open-set setting, i.e., unlabeled samples fall into novel classes, and proposes a class-aware consistency learning (CACL) method. First, to explore discriminative spectral-spatial features, a position-aware transformer is developed, which effectively models spatial position priors between the center pixel and its neighboring pixels via a symmetric position-aware encoding. Then, to reduce the interference from novel class samples on the model’s discrimination, a prototype-driven consistency learning is proposed, which accurately selects unlabeled samples belonging to known classes via a known class sampler, and efficiently utilizes their spectral-spatial information by modeling consistent predictions across different views. Finally, to further improve the distinguishability between known classes, a prototype contrastive optimization is proposed to decrease the distance between samples from the same class and increase the distances between those from different classes in the feature domain. Furthermore, an adaptive segmentation threshold is designed to accurately predict known classes and reject novel classes. Extensive experiments verify that our CACL outperforms the state-of-the-art methods, achieving the overall accuracy of 81.65%, 88.61%, and 92.88% on the Indian Pines, Salinas, and Pavia University datasets, with 10 labeled samples in each known class. The code is available at https://github.com/rock-in/CACL-main.

The methods of skeleton-based human action recognition (HAR) have gained widespread interest due to their robustness in capturing actions amidst changing environments and intricate backgrounds. Utilizing graph convolutional networks (GCNs) to describe the human skeleton for HAR has been shown to achieve impressive performance. However, most GCN-based methods only consider the relationships between adjacent joints, overlooking the relationships between joints that are not connected by natural physical links. Therefore, we propose a novel method called Multi-level Graph Convolutional Network, abbreviated as ML-GCN. We have refined the original partitioning strategy, introducing three novel strategies to more effectively enhance the connectivity among distant joints. Additionally, we incorporate a multi-level graph convolutional network and a non-local temporal convolutional network to better extract spatio-temporal features. Experiments conducted on the NTU-RGB+D and Kinetics datasets demonstrate that our model achieves a certain improvement in accuracy.

With rapid technological advancements in computer vision, the recognition of abnormal behavior during examinations has transitioned from human observation to computer-assisted recognition. Although traditional 2D Convolutional Neural Networks (CNNs) excel in computational efficiency, they need to capture crucial temporal dynamics for comprehensive video analysis more precisely. Nevertheless, 3D CNN-based methods demonstrate promising performance in temporal modeling but impose substantial computational demands and deployment costs. To overcome these challenges, this paper introduces an innovative Examination Anomaly Action Recognition Network named ReTANet. It incorporates cross-channel temporal modeling to capture temporal features within videos. It also employs Multi-Scale Channel Attention to enrich feature representation and extract channel and spatial information, thereby enhancing recognition accuracy without significantly increasing computational complexity and model parameters. Furthermore, this paper introduces the Examination Anomaly Action Dataset, also named the ExamGuard Dataset (EGD), to facilitate model training and evaluation. Remarkably, our model demonstrates superior performance compared to existing mainstream action recognition algorithms on the HMDB-5l dataset. Rigorous ablation studies conducted on the UCF-101 dataset have shown the effectiveness and significance of the proposed module.

Pseudo-label learning-based methods usually regard class confidence above a certain threshold for unlabeled samples as pseudo-labels, which may result in pseudo-labels still containing wrong labels. In this letter, we propose a prototype-based pseudo-label refinement (PPLR) for semi-supervised hyperspectral image classification. The proposed PPLR filters wrong labels from pseudo-labels using class prototypes, which can improve the discrimination of the network. First, PPLR uses multi-head attentions to extract the spectral-spatial features, and designs an adaptive threshold that can be dynamically adjusted to generate high-confidence pseudo-labels. Then, PPLR constructs class prototypes for different categories using labeled sample features and unlabeled sample features with refined pseudo-labels to improve the quality of pseudo-labels by filtering wrong labels. Finally, PPLR further assigns reliable weights to these pseudo-labels in calculating their supervised loss, and introduces a center loss to improve the discrimination of features. When 10 labeled samples per category are utilized for training, PPLR achieves the overall accuracies of 82.11%, 86.70% and 92.50% on the Indian Pines, Houston2013 and Salinas datasets, respectively.

Recently, pseudo-label-based deep learning methods have shown excellent performance in semi-supervised hyperspectral image (HSI) classification. These methods usually select high-confidence unlabeled samples to help optimize backbone classification networks. However, a large number of remaining low-confidence unlabeled samples, which contain rich land-covers information, are underutilized. In this paper, we propose a pseudo-label-based unreliable sample learning (PUSL) method to fully exploit low-confidence unlabeled samples for semi-supervised HSI classification. Firstly, to avoid overfitting the spatial distribution of labeled samples, we build a position-free transformer (PFT) as the backbone classification network. Secondly, PFT is initially trained with labeled samples in a supervised learning manner to obtain an initial classifier, which is then used to split unlabeled samples into reliable and unreliable unlabeled samples based on the predicted confidence. Thirdly, reliable unlabeled samples participate in training along with labeled samples. Finally, unreliable unlabeled samples are treated as negative samples for corresponding categories to improve the discrimination of PFT in a contrastive learning paradigm. Extensive experiments on three HSI datasets demonstrate that PUSL outperforms compared methods.

Accurately obtaining information of road traffic conditions is of great significance to people’s travel planning and arrangement of social shared resources, and has become a major research focus in the field of smart cities. Accurately predicting road conditions poses a huge challenge due to the complex spatial correlations and nonlinear temporal dependencies of real-time traffic networks. In this paper we propose a Multi-Range Attention-Based Dynamic Graph Convolutional Network (MRA-DGCN) to model complex traffic networks. The MRA-DGCN model uses a bicomponent modules to separate different periodicity to extract refined traffic signal. In the MRA-DGCN model, we use the adaptive spatial-temporal network block (ASTnet block), which includes dynamic graph convolution and temporal attention, to mine complex spatial correlations and nonlinear temporal dependencies, respectively. In the adaptive spatial-temporal network block, we use dynamically generated adjacency matrices instead of existing distance-based adjacency matrices to perform graph convolution operations to aggregate information between nodes during model training. Instead of hierarchically extracting spatial-temporal signal, we adopt temporal attention to capture the spatial-temporal information synchronously to improve the prediction performance. Furthermore, we propose a residual gated network to control the flow of information passed to the next hidden layer to enhance the predictive accuracy. Extensive experiments on two real-world traffic datasets, METR-LA and PeMS-BAY, show that the MRA-DGCN achieves the state-of-the-art results.

Accurate long-term traffic forecasting is of great significance to intelligent transportation systems, but long-term traffic forecasting is very challenging due to the complexity of the spatial and temporal relationship of traffic data. This paper proposes a model for prediction of traffic data. The model adopts an encoder-decoder architecture, in which the encoder and decoder are composed of a Graph Generator module, a Multi-Head Convolutional Self-Attention module, and a Dynamic Convolution Recurrent module to simulate the impact of spatial and temporal factors on traffic conditions. The encoder encodes the input features and the decoder predicts the output sequence. Between the encoder and the decoder, we use a Transformation mechanism to re-encode the traffic features to generate new sequence representations as the input of the decoder, which helps alleviate the problem of error propagation during prediction. We conduct the traffic prediction task on two real-world traffic datasets and the experimental results show that our model performs better than other baseline models.

In recent years, graph neural network has been widely used. Attention mechanism is introduced into the graph neural network to make it more applicable. Both GAT and AGNN prove that attention mechanism plays an important role in graph neural network. Attention mechanism algorithms such as gat and AGNN directly use a self-learning variable to do the point product after calculating the connection (or similarity calculation) of node and neighbor features (without further processing of the calculation results). Finally, we get an aggregation of neighbor information. A cosine similarity distance pruning algorithm based on graph attention mechanism (CDP-GA) is proposed to optimize the attention matrix of nodes and their adjacent nodes. By calculating the cosine similarity between node features and neighbor features (the feature here is obtained by linear transformation), the similarity of nodes is regarded as the distance between nodes (or the weight of edges). And we think that the aggregation degree of node information is inversely proportional to the distance between nodes (similar to the heat conduction formula). In the method, we prune the neighborhood of the node according to the cosine similarity to get the final attention coefficient matrix. In this way, the attention mechanism in the graph neural network is further refined, and the loss of aggregation neighbor information is reduced. In the experiments of three datasets, our model is compared with the experimental classification of GAT and AGNN and the experiment of correlation graph neural network algorithm. Finally, it is proved that the algorithm is better than three known datasets.

The problem of session-based recommendation model aims to predict user actions based on anonymous sessions. Although, previous models achieved promising results, there are still some problems, for example, we are unable to take into account the effects of session sequences of different lengths. Generally speaking, the effect of long sequence is not as good as that of short sequence in the same model. The reason of above is that the characteristics of different length session will vary greatly. Generally, the shorter the session, the tighter the relationship between items, and the longer the session, the more likely there are items that have no relationship with each other. So, we propose a model named session-based recommendation model based on multiple neural networks hybrid extraction feature. This model uses different feature extractor to deal with the features of long sessions and short sessions respectively. In SR-MNN, we use Graph Convolutional Network to extract the features of long session and use Recurrent Neural Network to extract the features of short session. Each session is then represented as the composition of the global preference, the initial interest of that session, and the current interest of that session using an attention network. Experiments on two real datasets show that SR-MNN evidently outperforms the state-of-the-art session-based recommendation methods consistently.

In protein-protein interaction networks, proteins combine into macromolecular complexes to execute essential functions in the cells, such as replication, transcription, protein transport. Considering the certain rate of false positive and false negative interactions, we take a confidence probability on interactions and correlate interactions and gene expression data to assign weights to edges in PPI networks. Then we propose the CIGE algorithm to detecting protein complexes from protein interaction networks. Our algorithm takes a maximal full-connected sub-graph as core graph of a seed node, and decides whether a node belongs to a protein complex through judging in-module weight and out-module weight between core graph and nodes out of core graph. Experiment results show that our algorithm has an excellent performance in both accuracy and hit rate.

This paper addresses dynamic restoration to the scenario of single link failure for survivable traffic grooming in WDM mesh networks. Two restoration schemes are proposed, dynamic restoration at connection granularity (DRAC) and dynamic restoration at lightpath granularity (DRAL). DRAC routes a failed connection to a new multihop path, while DRAL routes a failed lightpath to a new lightpath or other available lightpath. Then we compare the two restoration schemes with other schemes for survivable traffic grooming.

This paper addresses the dynamic traffic grooming problem in wavelength-division-multiplexed (WDM) mesh optical networks subject to wavelength constraint and transceiver constraint. As an improvement over the existing link bundled auxiliary graph (LBAG) model which thinks all traffic streams must go though the grooming fabric before entering the wavelength switch fabric, we introduce a transceiver saving auxiliary graph (TSAG) model and propose the TSAG method. In the TSAG model, we can determine whether a traffic stream should consume grooming ports or not. Various grooming policies are achieved by assigning the weight value of different edges in the auxiliary graph and their blocking performance are compared through simulations. Results show that the TSAG model has a lower blocking probability than LBAG while consuming the less running time, and TSAG can reduce the number of transceivers utilized while improving the wavelength utilization