Malware can greatly compromise the integrity and trustworthiness of information and is in a constant state of evolution. Existing feature fusion‐based detection methods generally overlook the correlation between features. And mere concatenation of features will reduce the model’s characterization ability, lead to low detection accuracy. Moreover, these methods are susceptible to concept drift and significant degradation of the model. To address those challenges, we introduce a feature graph‐based malware detection method, malware feature graph (MFGraph), to characterize applications by learning feature‐to‐feature relationships to achieve improved detection accuracy while mitigating the impact of concept drift. In MFGraph, we construct a feature graph using static features extracted from binary PE files, then apply a deep graph convolutional network to learn the representation of the feature graph. Finally, we employ the representation vectors obtained from the output of a three‐layer perceptron to differentiate between benign and malicious software. We evaluated our method on the EMBER dataset, and the experimental results demonstrate that it achieves an AUC score of 0.98756 on the malware detection task, outperforming other baseline models. Furthermore, the AUC score of MFGraph decreases by only 5.884% in 1 year, indicating that it is the least affected by concept drift.

In the current digital age, users store their personal information in corporate databases to access services, making data security and sensitive information protection central to enterprise security management. Given the extensive attack surface, system assets continuously face cyber security challenges, such as weak authentication, exploitation of system vulnerabilities, and malicious software. Through specific vulnerabilities, attackers may gain unauthorized system access, masquerading as legitimate users, and remaining hidden. Successful attacks can lead to the leakage of user privacy, disruption of business operations, significant financial losses, and damage to corporate reputation. The increasing complexity of attack vectors is blurring the boundaries between insider and external threats. To address this issue, this article introduces the IDU-Detector, an innovative threat detection framework that strategically integrates intrusion detection systems (IDSs) with user and entity behavior analytics (UEBA). This integration aims to monitor unauthorized access and malicious attacks within systems, bridging functional gaps between existing systems, ensuring continuous monitoring and real-time response of the network environment, and enhancing their collective effectiveness in identifying security threats. Additionally, the existing insider threat datasets exhibit significant deficiencies in both depth and comprehensiveness, lacking sufficient coverage of diverse attack vectors. This limitation hinders the ability of insider threat detection technologies to effectively address the growing complexity and expanding scope of sophisticated attack surfaces. To address these gaps, we propose new, more enriched and diverse datasets that includes a wider range of attack scenarios, thereby enhancing the adaptability and effectiveness of detection technologies in complex threat environments. We tested our framework on different datasets, the IDU-Detector achieved average accuracy rates of 98.96% and 99.12%. These results demonstrate the method’s effectiveness in detecting masquerader attacks and other malicious activities, significantly improving security protection and incident response speed, and providing a higher level of security assurance for asset safety.

Malware is proliferating at an exponential rate in cyberspace, posing serious threats to on-device systems characterized by limited computational capabilities. In this work, we address the critical challenge posed by data imbalance—where rare malware families receive inadequate representation—by proposing MalViT, a lightweight Vision Transformer (ViT) architecture that directly operates in the JPEG frequency domain. Rather than converting Huffman-coded signals into RGB spatial images, MalViT leverages Discrete Cosine Transform (DCT) coefficients to reduce data redundancy and computational overhead. We further improve the model’s generalization through both pre-training and fine-tuning workflows. Comprehensive evaluations on two large-scale, real-world malware datasets, MalNet-Image (1.26 M samples) and BODMAS (51 K samples), demonstrate that MalViT accelerates data loading by nearly threefold compared to existing methods. On GPU and CPU, MalViT achieves approximately 2.0× and 4.7× faster inference throughput than MobileViT, respectively, while incurring minimal or even improved accuracy loss. When processing 224 × 224-pixel JPEG images, MalViT completes inference within an average of 3.12ms per sample, which is 8.79× faster than VisMal and 5.91× faster than ViT4Mal. Furthermore, its compact design comprises only 1.1M parameters and requires 10M MACs, making it particularly suitable for resource-constrained on-device deployment.