Deep neural networks are widely applied in computer vision tasks but remain highly vulnerable to adversarial attacks. Tiny and imperceptible perturbations can cause severe model misclassification. Most existing defense methods improve robustness but significantly reduce model accuracy on clean examples. To address this issue, we propose a defense framework combining pixel value transformation and spatial transformation. The proposed method divides the input image into two complementary regions. Feature compression is applied to one region to reduce model sensitivity to subtle perturbations. Intense reversible pixel transformation is applied to the other region to disrupt the spatial distribution of the perturbations. Furthermore, a spatial transformation module is introduced to enhance defense capabilities against strong attacks. This module adjusts the spatial structure of the image, increases nonlinear relationships among pixels, and suppresses perturbation transfer effects. A Multiple Objective Particle Swarm Optimization (MOPSO) is employed to adaptively search for optimal transformation parameters, achieving a balance between robustness and accuracy. An inverse transformation mechanism is proposed to restore attacked examples back to a form recognizable to human eyes, which effectively destroys the imperceptible nature of adversarial examples. Experimental results demonstrate that the proposed method successfully resists various adversarial attacks while maintaining high recognition accuracy on clean images. The proposed reversible defense framework demonstrates theoretical universality and scalability, providing an effective new approach for adversarial defense.

With the popularity of social networks, social bots are increasingly interfering with human social activities, which have negatively impacted social network ecology and user experience. How to detect social bots effectively has become a challenging task. Most of existing social bot detection methods are based on the supervised learning mechanism, but the labelled data collection is resource-intensive. In this paper, a Multi-stage Selftraining Social Bot detection method based on a Graph Neural Network (MSSBot) is proposed to improve the performance of social bot detection with a small amount of labelled data. MSSBot exploits pseudo-labels to expand the dataset and trains a Relational Graph Convolutional Network in multiple stages, while uncertaintyaware and deep clustering are used to reduce the noise present in the pseudo-labels acquisition process. The experiments were conducted on a real-world social network dataset Twibot-20, a small high-quality dataset focused on social bot behavior patterns Cresci-2015, and a large-scale graph-based dataset Twibot-22 with high annotation quality. The experimental results show that our method achieves accuracies of 0.9259, 0.9979, and 0.8358 on the Twibot-20, Cresci-2015, and Twibot-22 datasets, respectively, outperforming the state-of-the-art methods, with improvements of 5.68 %, 2.64 %, and 3.92 % over the existing state-of-the-art methods.

Low rank adaptation (LoRA) methods have demonstrated strong capabilities in efficiently fine-tuning large models. However, existing LoRA-based approaches typically require manually setting the scaling factor, a process that involves extensive search efforts to find optimal values. To address this challenge, we first develop data-driven heuristic methods that automatically determine layer-wise scaling factors through either activation pattern analysis during forward propagation or gradient behavior monitoring during backward updates. However,their practical performance remains unsatisfactory in applications. Building upon these theoretical foundations, we present MSLoRA, a novel framework that reformulates scaling factor determination as a dynamic optimization problem in parameter-efficient fine-tuning. Our approach innovatively models scaling factors as self-adaptive meta-parameters whose optimal values emerge organically through the interplay between transformer architecture hierarchies and task-specific learning objectives. Extensive experiments conducted across both natural language understanding and generative tasks reveal that MSLoRA consistently outperforms baseline models. This highlights the effectiveness of MSLoRA's dynamic, layer-specific adjustment mechanism in capturing the complex nature of task-specific activation patterns, making it a more robust and scalable solution for parameter-efficient fine-tuning of large models.

Traditional ensemble methods rely on majority voting, which may fail to recognize correct answers held by a minority in scenarios requiring specialized knowledge. Therefore, this paper proposes two novel ensemble methods for supervised classification, named Confidence Truth Serum (CTS) and Confidence Truth Serum with Single Regression (CTS-SR). The former is based on the principles of Bayesian Truth Serum (BTS) and introduces classification confidence to calculate the prior and posterior probabilities of events, enabling the recovery of correct judgments provided by a confident minority beyond majority voting. CTS-SR further simplifies the algorithm by constructing a single regression model to reduce computational overhead, making it suitable for large-scale applications. Experiments are conducted on multiple binary classification datasets to evaluate CTS and CTS-SR. Experimental results demonstrate that, compared with existing ensemble methods, both of the proposed methods significantly outperform baseline algorithms in terms of accuracy and F1 scores. Specifically, there is an average improvement of 2–6% in accuracy and an average increase of 2–4% in F1 score. Notably, on the Musk and Hilly datasets, our method achieves a 5% improvement compared to the traditional majority voting approach. Particularly on the Hilly dataset, which generally exhibits the poorest classification performance and poses the greatest prediction challenges, our method demonstrates the best discriminative performance. validating the importance of confidence as a feature in ensemble learning.

Electroencephalogram (EEG)-based emotion recognition has become a focus of brain-computer interface research. However, differences in EEG signals across subjects can lead to poor generalization. Moreover, current approaches individually extract temporal and spatial information, resulting in inadequate feature fusion during feature extraction. This study develops a novel ChannelMix-based transformer and convolutional multi-view feature fusion network (CMTCF) to enhance cross-subject EEG emotion recognition. Specifically, a bidirectional fusion module based on a convolutional neural network (CNN)-Transformer structure is introduced to extract multi-view spatial feature and temporal feature, enabling the representation of rich spatiotemporal information. Subsequently, the ChannelMix module is designed to effectively establish an intermediate domain, facilitating the alignment of the target and source domains to reduce their discrepancies. Additionally, a soft pseudo-label module is implemented to enhance the discriminative power of target domain data within the feature space. To further improve generalization, a ChannelMix-based data augmentation method is utilized. Comprehensive experiments are conducted on the SEED, SEED-IV and SEED-VII benchmark datasets, achieving recognition accuracies of 93.80% (+/- 4.96), 79.37% (+/- 6.05) and 49.13% (+/- 8.22), respectively, demonstrating that the CMTCF network achieves competitive results in cross-subject EEG emotion recognition tasks.

The proliferation of malicious social bots poses severe threats to cybersecurity and social media information ecosystems. Existing detection methods often overlook the semantic value and emotional cues conveyed by emojis in user-generated tweets. To address this gap, we propose ESA-BotRGCN, an emoji-driven multi-modal detection framework that integrates semantic enhancement, sentiment analysis, and multi-dimensional feature modeling. Specifically, we first establish emoji–text mapping relationships using the Emoji Library, leverage GPT-4 to improve textual coherence, and generate tweet embeddings via RoBERTa. Subsequently, seven sentiment-based features are extracted to quantify statistical disparities in emotional expression patterns between bot and human accounts. An attention gating mechanism is further designed to dynamically fuse these sentiment features with user description, tweet content, numerical attributes, and categorical features. Finally, a Relational Graph Convolutional Network (RGCN) is employed to model heterogeneous social topology for robust bot detection. Experimental results on the TwiBot-20 benchmark dataset demonstrate that our method achieves a superior accuracy of 87.46%, significantly outperforming baseline models and validating the effectiveness of emoji-driven semantic and sentiment enhancement strategies.

With the development of deep learning technology, convolutional neural network (CNN) has been widely used in many fields such as face recognition, automatic driving, biomedicine, etc., replacing human beings to complete complex and redundant work, which brings great convenience to people’s lives. However, the discovery and development of adversarial examples have created a greater threat to image recognition. In this paper, we propose an adaptive image adversarial example detection method based on class activation mapping, which utilizes the hot zone discovery results of the Grad-CAM algorithm to perform adaptive noise reduction on images and analyzes the differences in the classification results of images before and after the noise reduction in the same benchmark network, including the KL dispersion, the label change, the label confidence, etc., to achieve the detection of adversarial examples on the ImageNet-1000 dataset. The experimental results show that the algorithm proposed in this paper achieves better detection results, and the F1 reaches 0.82 in detecting the generated FGSM adversarial examples with ϵ = 0.3, which is better than the baseline model.

The emergence of malicious Twitter social bots poses a considerable threat to the security of social networks, and the detection of evolving social bots has become challenging. State-of-the-art detection methods are usually supervised, but the label acquisition process suffers from time-consuming and inaccurate problems in addition to its inability to cope with the challenge of the continuous evolution of social bots. Moreover, structural relationships within social networks are under-explored using current unsupervised methods. To address current challenges, we propose an unsupervised social bot detection method based on deep contrastive graph clustering (BotDCGC). This approach utilises a graph attentional encoder and an inner product decoder to acquire user node embeddings. By incorporating information from both user account features and topological structure, the model employs a contrastive learning technique based on structure to discern node embeddings of distinct classes within the feature space. Subsequently, confident cluster assignments are used as soft labels to guide the embedding process by calculating the similarity between each node and the clustering centre, enabling the joint optimisation of user node embeddings and clustering results. Experiments show that BotDCGC is more effective than the state-of-the-art baselines with an accuracy of 0.8095 in the Twibot-20 dataset and 0.9334 in the Cresci-2015 dataset, and the experimental results indicate the effectiveness of the graph autoencoder module, contrastive learning and deep clustering module.

Malicious social bots pose a serious threat to social network security by spreading false information and guiding bad opinions in social networks. The singularity and scarcity of single organization data and the high cost of labeling social bots have given rise to the construction of federated models that combine federated learning with social bot detection. In this paper, we first combine the federated learning framework with the Relational Graph Convolutional Neural Network (RGCN) model to achieve federated social bot detection. A class-level cross entropy loss function is applied in the local model training to mitigate the effects of the class imbalance problem in local data. To address the data heterogeneity issue from multiple participants, we optimize the classical federated learning algorithm by applying knowledge distillation methods. Specifically, we adjust the client-side and server-side models separately: training a global generator to generate pseudo-samples based on the local data distribution knowledge to correct the optimization direction of client-side classification models, and integrating client-side classification models’ knowledge on the server side to guide the training of the global classification model. We conduct extensive experiments on widely used datasets, and the results demonstrate the effectiveness of our approach in social bot detection in heterogeneous data scenarios. Compared to baseline methods, our approach achieves a nearly 3–10% improvement in detection accuracy when the data heterogeneity is larger. Additionally, our method achieves the specified accuracy with minimal communication rounds.

Black-box attacks generate adversarial examples by querying the target model and updating the noise according to the feedback. However, the current black-box attack methods require excessive queries to generate adversarial examples, increasing the risk of detection by target defense systems. Furthermore, the current black-box attack methods primarily focus on controlling the magnitude of perturbations while neglecting the impact of perturbation placement on the stealthiness of adversarial examples. To this end, we propose a novel edge noise-constrained black-box attack method using the artificial fish swarm algorithm (EFSAttack). EFSAttack introduces the concept of edge noise constraint to indicate the low-frequency region of the image where perturbations are added and employs edge noise constraint to improve the population initialization and population evolution process. The experiments on CIFAR-10 and MNIST show notable improvements in the success rates, query efficiency, and adversarial example invisibility.

The number of social hots on online social networks has increased significantly. Although social media platforms and researchers have taken sonic measures to address this issue, most of the methods have riot focused on the relationships in social networks. The popular undirected graphs used in current times not only fail to express the diverse relationships between users, but are also limited by the single neighbor relationship. This paper presents a new graph model called E-GraphRTAGE, which can process different types of user relationships while not neglecting the role of its own nodes in the relationship. Additionally, it expands the scope of graph models by incorporating CNN's convolutional kernels, no longer focusing solely on neighboring nodes but also paying attention to distant nodes or nodes at a greater distance. We input the Twi-bot20 dataset into our model for social bot detection and achieved good results.

ASCII art detection and recognition is an important branch of current network information processing. However, due to ASCII art's text-based organization and image-based semantic expression, traditional natural language processing (NLP) and image recognition fail to yield ideal results. This paper designs an ASCII art localization and extraction algorithm based on string distance for highly mixed text and ASCII art, aiming to segment clean ASCII art for subsequent recognition. Additionally, an evaluation standard for ASCII art extraction effectiveness is defined. Experimental results show that the proposed algorithm performs well in locating and extracting ASCII art.

The number of social bots in OSNs(Online Social Networks) ,has increased significantly, which brings great risks to social network security. Aiming at the problem of insufficient emoji usage in the existing social bot detection, this paper proposes a social account representation method combining emoji information and is used for social bot detection. The emojiswitch library was used to replace the text of emoji, and the word embedding model RoBERTa was used to obtain the tweet features. User description, user tweets, user numerical properties, user categorical properties and user relationship are combined to realize the complete emoji of social accounts. The effectiveness of the proposed method is verified based on the public data set.

Feature selection reduces the dimensionality of high-dimensional data by removing redundant or irrelevant features from the original features, thus reducing the negative impact of the "dimensionality curse." Subspace clustering feature selection methods focus on the structure and properties within the dataset, so they perform well in unsupervised feature selection work. In this paper, we sort out and classify the research on subspace clustering feature selection and propose several future research trends based on the current status of feature selection in subspace clustering.

Popular topics in the field of countermeasures include feature engineering and neural-network-based models, which involve neural network architectures and loss criteria. This study focuses on Res2Net and its variant models to examine the impact of model generalization on countermeasure performance in the ASVspoof 2019 logical access and physical access scenarios. Results reveal that while Res2Net exhibits superior generalization compared to its variants, the most effective countermeasure combines both feature engineering and model optimization. The proposed dynamic modulated-Res2Net utilizes channel-wise soft attention to recalibrate feature maps, offering adaptive adjustments to spoofing cues of varying scales. Evaluation on the logical access dataset demonstrates dynamic modulated-Res2Net's relative improvement of over 38% compared to Res2Net. Furthermore, we exploit low-frequency features and combine them with dynamic modulated-Res2Net to achieve in an equal error rate of 1.21% under logical access and 0.41% under physical access, establishing our proposed dynamic modulated-Res2Net as one of the top-performing single systems. Additionally, we compare the best countermeasures in different scenarios, highlighting the ongoing challenge of achieving generalization.

针对网络安全课程知识点分散且难度大,所学知识不能系统理解、无法举一反三的问题,提出将成语及"三十六计"为代表的优秀传统文化融入教学环节,介绍具体教学应用过程,最后以实际教学数据说明教学应用效果.

Social engineering attacks are considered one of the most hazardous cyberattacks in cybersecurity, as human vulnerabilities are often the weakest link in the entire network. Such vulnerabilities are becoming increasingly susceptible to network security risks. Addressing the social engineering attack defense problem has been the focus of many studies. However, two main challenges hinder its successful resolution. Firstly, the vulnerabilities in social engineering attacks are unique due to multistage attacks, leading to incorrect social engineering defense strategies. Secondly, social engineering attacks are real-time, and the defense strategy algorithms based on gaming or reinforcement learning are too complex to make rapid decisions. This paper proposes a multiattribute quantitative incentive method based on human vulnerability and an improved Q-learning (IQL) reinforcement learning method on human vulnerability attributes. The proposed algorithm aims to address the two main challenges in social engineering attack defense by using a multiattribute incentive method based on human vulnerability to determine the optimal defense strategy. Furthermore, the IQL reinforcement learning method facilitates rapid decision-making during real-time attacks. The experimental results demonstrate that the proposed algorithm outperforms the traditional Q-learning (QL) and deep Q-network (DQN) approaches in terms of time efficiency, taking 9.1% and 19.4% less time, respectively. Moreover, the proposed algorithm effectively addresses the non-uniformity of vulnerabilities in social engineering attacks and provides a reliable defense strategy based on human vulnerability attributes. This study contributes to advancing social engineering attack defense by introducing an effective and efficient method for addressing the vulnerabilities of human factors in the cybersecurity domain.

Rapid advancement of intelligent transportation systems (ITS) and autonomous driving (AD) have shown the importance of accurate and efficient detection of traffic signs. However, certain drawbacks, such as balancing accuracy and real-time performance, hinder the deployment of traffic sign detection algorithms in ITS and AD domains. In this study, a novel traffic sign detection algorithm was proposed based on the bidirectional Res2Net architecture to achieve an improved balance between accuracy and speed. An enhanced backbone network module, called C2Net, which uses an upgraded bidirectional Res2Net, was introduced to mitigate information loss in the feature extraction process and to achieve information complementarity. Furthermore, a squeeze-and-excitation attention mechanism was incorporated within the channel attention of the architecture to perform channel-level feature correction on the input feature map, which effectively retains valuable features while removing non-essential features. A series of ablation experiments were conducted to validate the efficacy of the proposed methodology. The performance was evaluated using two distinct datasets: the Tsinghua-Tencent 100K and the CSUST Chinese traffic sign detection benchmark 2021. On the TT100K dataset, the method achieves precision, recall, and Map0.5 scores of 83.3%, 79.3%, and 84.2%, respectively. Similarly, on the CCTSDB 2021 dataset, the method achieves precision, recall, and Map0.5 scores of 91.49%, 73.79%, and 81.03%, respectively. Experimental results revealed that the proposed method had superior performance compared to conventional models, which includes the faster region-based convolutional neural network, single shot multibox detector, and you only look once version 5.

With the development of science and computer technology, social networks are changing our daily lives. However, this leads to new, often hidden dangers in areas such as cybersecurity. Of these, the most complex and harmful is the Advanced Persistent Threat attack (APT attack). The development of personality analysis and prediction technology provides the APT attack a good opportunity to infiltrate personality privacy. Malicious people can exploit existing personality classifiers to attack social texts and steal users’ personal information. Therefore, it is of high importance to hide personal privacy information in social texts. Based on the personality privacy protection technology of adversarial examples, we proposed a Supervised Character Resemble Substitution personality adversarial method (SCRS) in this paper, which hides personality information in social texts through adversarial examples to realize personality privacy protection. The adversarial examples should be capable of successfully disturbing the personality classifier while maintaining the original semantics without reducing human readability. Therefore, this paper proposes a measure index of “label contribution” to select the words that are important to the label. At the same time, in order to maintain higher readability, this paper uses character-level resemble substitution to generate adversarial examples. Experimental validation shows that our method is able to generate adversarial examples with good attack effect and high readability.

Feature selection has become a vital issue in data mining and machine learning. But some challenges have been outstanding when trying to improve the performance of feature selection, such as small sample, uncertain classes, complex features, complementation and redundancy between each feature. In this paper, firstly the background of feature selection is introduced. Then we have presented a new perspective to analyze multi-label feature selection and provided typical papers on different classifications. To further analyze these algorithms, evaluation criterion on results of multi-label feature selection is summarized. Finally, some reflects on research directions, future works and conclusions are organized.