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.

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.

Deep hashing is frequently utilized in large-scale image retrieval because of its strong representation learning capabilities and effective processing capacity. However, deep hashing models are susceptible to adversarial examples. We propose a label-guided weighted autoencoder network (LGWAE) to generate adversarial examples for targeted hashing attacks. Specifically, we first introduce a multi-label learning network to extract the semantic features and the category code of different labels. Then, the semantic features and the benign image are collectively fed into an autoencoder in order to generate a natural-looking adversarial example. The category code is used as target hash code to supervise the generation of the adversarial example. To efficiently generate better-performing adversarial examples, we design a weighted Hamming distance loss that dynamically adjusts the loss value based on the label similarity between the benign image label and the target label. Numerous experiments demonstrate that we are capable of efficiently and effectively generating better quality adversarial samples.