Lei Xiao

Domain Generation Algorithm(DGA) is widely used by botnets to evade detection by generating numerous pseudo-random domains to communicate with commandandcontrol servers. While existing Graph Neural Networks attempt to detect DGA botnets by exploiting the feature similarity of these domains to model semantic associations via similarity graphs, they are restricted to binary relationships, causing information decay during multi-hop propagation. To overcome this, we propose HyperDGA. Treating domains as nodes, HyperDGA utilizes K Nearest Neighbors to construct hyperedges, explicitly capturing high order group semantic correlations. Subsequently, a Local Topology Aggregation module employs multi-head node attention-based hypergraph convolution to dynamically assign distinct aggregation weights to intra hyperedge nodes, extracting fine-grained structural features. To mitigate the limited receptive field of hypergraph convolutions, a Global Node Association module integrates the selective state space model, Mamba, to capture long-range dependencies across all nodes. Experiments on two public datasets demonstrate that HyperDGA outperforms all baselines and achieves over 99% accuracy, validating the efficacy of high-order semantic modeling for DGA botnet detection.

As network automation and self-organizing networks (SONs) rapidly evolve, edge devices increasingly demand lightweight, real-time, and high-precision classification algorithms to support critical tasks such as traffic identification, intrusion detection, and fault diagnosis. In recent years, cross-entropy (CE) loss has been widely adopted in deep learning classification tasks due to its computational efficiency and ease of optimization. However, traditional CE methods primarily focus on class separability without explicitly constraining intra-class compactness and inter-class boundaries in the feature space, thereby limiting their generalization performance on complex classification tasks. To address this issue, we propose a novel classification loss framework—Contrastive Geometric Cross-Entropy (CGCE). Without incurring additional computational or memory overhead, CGCE explicitly introduces learnable class representation vectors and constructs the loss function based on the dot-product similarity between features and these class representations, thus explicitly reinforcing geometric constraints in the feature space. This mechanism effectively enhances intra-class compactness and inter-class separability. Theoretical analysis further demonstrates that minimizing the CGCE loss naturally induces clear and measurable geometric class boundaries in the feature space, a desirable property absent from traditional CE methods. Furthermore, CGCE can seamlessly incorporate the prior knowledge of pretrained models, converging rapidly within only a few training epochs (for example, on the CIFAR-10 dataset using the ViT model, a single training epoch is sufficient to reach 99% of the final training accuracy.) Experimental results on both text and image classification tasks show that CGCE achieves accuracy improvements of up to 2% over traditional CE methods, exhibiting stronger generalization capabilities under challenging scenarios such as class imbalance, few-shot learning, and noisy labels. These findings indicate that CGCE has significant potential as a superior alternative to traditional CE methods.

PurposeMedical image segmentation based on deep learning has played a crucial role in computer-aided medical diagnosis. However, they are still vulnerable to imperceptible adversarial attacks, which lead to potential misdiagnosis in clinical practice. Research on adversarial attack methods is beneficial for improving the robustness design of medical image segmentation models. Currently, there is a lack of research on adversarial attack methods toward deep learning-based medical image segmentation models. Existing attack methods often yield poor results in terms of both attack effects and image quality of adversarial examples and primarily focus on nontargeted attacks. To address these limitations and further investigate adversarial attacks on segmentation models, we propose an adversarial attack approach.ApproachWe propose an approach called momentum-driven adaptive feature-cosine-similarity with low-frequency constraint attack (MAFL-Attack). The proposed feature-cosine-similarity loss uses high-level abstract semantic information to interfere with the understanding of models about adversarial examples. The low-frequency component constraint ensures the imperceptibility of adversarial examples by constraining the low-frequency components. In addition, the momentum and dynamic step-size calculator are used to enhance the attack process.ResultsExperimental results demonstrate that MAFL-Attack generates adversarial examples with superior targeted attack effects compared with the existing Adaptive Segmentation Mask Attack method, in terms of the evaluation metrics of Intersection over Union, accuracy, L2, L∞, Peak Signal to Noise Ratio, and Structure Similarity Index Measure.ConclusionsThe design idea of the MAFL-Attack inspires researchers to take corresponding defensive measures to strengthen the robustness of segmentation models.

Filtering high-variability and high-severity defect reports from large test report databases is a challenging task in crowdtesting. Traditional optimization algorithms based on clustering and distance techniques have made progress but are limited by initial parameter settings and significantly decrease in efficiency with an increasing number of reports. To address this issue, this paper proposes a method that integrates reinforcement learning with genetic algorithms for crowdsourced testing report optimization, called Reinforcement Learning-based Genetic Algorithm for Crowdsourced Testing Report Optimization (RLGA). Its core goal is to identify distinct, high-severity defect reports from a large set. The method uses genetic algorithms to generate the optimal report selection sequence and adjusts the crossover probability (Pc) and mutation probability (Pm) dynamically with reinforcement learning based on the population’s average fitness, best fitness, and diversity. The reinforcement learning component uses a hybrid SARSA and Q-Learning strategy to update the Q-value table, allowing the algorithm to learn quickly in early iterations and expand the search space later to avoid local optima, thereby improving efficiency. To validate the RLGA method, this paper uses four public datasets and compares RLGA with six classic methods. The results indicate that RLGA outperforms BDDIV in terms of execution time and is less sensitive to the total number of test reports. In terms of optimization objectives, the test reports selected by RLGA have higher levels of defect severity and diversity than those selected by the random choice, BDDIV, and TSE methods. Regarding population diversity, RLGA effectively enhances the uniformity and diversity of individuals compared to random initialization. In terms of convergence speed, RLGA is superior to the GA, GA-SARSA, and GA-Q methods.

Code classification and code clone detection are crucial for understanding and maintaining large software systems. Although deep learning surpasses traditional techniques in capturing the features of source code, existing models suffer from low processing power and high complexity. We propose a novel source code representation method based on the multi-head attention mechanism (SCRMHA). SCRMHA captures the vector representation of entire code segments, enabling it to focus on different positions of the input sequence, capture richer semantic information, and simultaneously process different aspects and relationships of the sequence. Moreover, it can calculate multiple attention heads in parallel, speeding up the computational process. We evaluate SCRMHA on both the standard dataset and an actual industrial dataset, and analyze the differences between these two datasets. Experiment results in code classification and clone detection tasks show that SCRMHA consumes less time and reduces complexity by about one-third compared with traditional source code feature representation methods. The results demonstrate that SCRMHA reduces the computational complexity and time consumption of the model while maintaining accuracy.

Crowdsourced testing for mobile applications is widespread. Efficiently and swiftly inspecting enormous test reports is a crucial challenge in crowdsourced testing. Most existing methods focus on finding different bugs earlier, i.e. the diversity of bugs, while ignoring bug severity. We present a multi-objective prioritization method for crowdsourced test reports that takes into account the diversity and severity of bugs. In the beginning, we ranked the test reports according to severity, then clustered the test reports by fusing textual and image features, and finally used the clustering results to adjust the initial ranking to ensure diversity. To validate our method, we conducted experiments using an industrial crowdsourced test report dataset comprised of six mobile application projects. The results demonstrate that our method can find more different high-severity bugs earlier than existing methods.

Collating vast test reports is a time-consuming and laborious task in crowdsourced testing. Crowdsourced test reports are usually presented in two ways, one as text and the other as images, which have symmetrical content. Researchers have proposed many text- and image-based methods for prioritizing crowdsourced test reports of mobile applications. However, crowdsourced test reports of web applications typically have clearer textual descriptions of errors but noisier screenshots of errors. This gap motivates us to propose a method for prioritizing crowdsourced test reports of web applications to detect all errors earlier. In this paper, we integrate text and image information from test reports to enhance the analysis process. First, we use the natural language processing (NLP) technique to extract textual features of error descriptions and then symmetrically extract image features of error screenshots, i.e., we use the optical character recognition (OCR) technique to obtain textual information in the screenshots and then also use the NLP technique to extract features. To validate our approach, we conduct experiments on 717 test reports. The experimental results show that our method has a higher APFD (average percentage fault detection) and shorter runtime than state-of-the-art prioritization methods.

In crowdsourced testing, prioritizing numerous test reports is critical for improving developer review efficiency. Many researchers have proposed methods for prioritizing crowdsourced test reports for mobile applications, however, web crowdsourced test reports usually contain more detailed text and more complex screenshots, which makes prioritizing them especially necessary. This paper proposes a prioritization method named TCDiv, based on the data characteristics of web crowdsourcing test reports. First, using a pre-trained TextCNN model, the text of each report is segmented into two parts: reproduction steps and defect descriptions. Then, features are extracted from textual and image information respectively, and clustering technique is utilized to classify similar reports, and finally the clustering results are ranked and sampled. To validate the approach, experiments were conducted on 717 web crowdsourced test reports. The results show that TCDiv can detect more different defects in a limited time, thus improving the review efficiency of the development team.

Crowdsourced testing has been widely used to improve software quality since it can quickly obtain a vast number of test reports revealing various bugs. One of the most critical tasks in crowdsourced testing is determining how to inspect these valuable reports efficiently and quickly. In recent years, many automated techniques, like clustering and prioritizing, have arisen to assist in examining crowdsourced test reports. Due to the high repetition rate, most existing crowdsourced test report prioritization techniques concentrate on finding different bugs earlier while ignoring bug severity. The bug's severity reflects how detrimental the bug is to the quality of the software. In practice, developers should fix highly severe bugs as early as possible. In this paper, we propose a multi-objective prioritization technique for crowdsourced test reports that considers the diversity and severity of bugs. First, test reports are sorted in order of severity, and then the clustering results of test reports are used to adjust the first sorting sequence to ensure diversity. To acquire the clustering results, we first obtain the text feature vector and image feature vector of the test report separately, then fuse these two feature vectors. Finally, we use the density clustering algorithm to cluster the fused feature vectors. We conducted experiments on an industrial crowdsourced test report dataset from six mobile application projects to validate our method. The results show that our method can detect serious bugs earlier and different bugs faster than existing methods.