A5100680264- Advanced Malware Detection Techniques
- Network Security and Intrusion Detection
- Digital and Cyber Forensics
- Service-Oriented Architecture and Web Services
- Cell Image Analysis Techniques
- Anomaly Detection Techniques and Applications
- Advanced Decision-Making Techniques
- Semantic Web and Ontologies
- Artificial Immune Systems Applications
- Advanced Measurement and Detection Methods
- Data Mining Algorithms and Applications
- Bacillus and Francisella bacterial research
- Web Data Mining and Analysis
- Military Defense Systems Analysis
- Multi-Agent Systems and Negotiation
181st Hospital of Chinese People's Liberation Army
2024
Air Force Engineering University
2005-2022
Inner Mongolia University of Technology
2011
Abstract With the advancement of adversarial techniques for malicious code, malevolent attackers have propagated numerous code variants through shell coding and obfuscation. Addressing current issues insufficient accuracy efficiency in classification methods based on deep learning, this paper introduces a detection strategy uniting Convolutional Neural Networks (CNNs) Transformers. This approach utilizes neural architecture, incorporating novel fusion module to reparametrize structure, which...
Support Vector Machine (SVM) is a relatively novel classification technology, which has shown higher performance than traditional learning methods in many applications. Therefore, some security researchers have proposed an intrusion detection method based on SVM. However, the SVM algorithm very sensitive to choice of kernel function and parameter adjustment. Once selection unscientific, it will lead poor accuracy. To solve this problem, paper presents Grey Wolf Optimizer Algorithm Particle...
According to the Web log mining, site administrators can control network traffic and understand user access modes. Then they further improve performance of systems optimize system design sites by using these information. However, data doesn't perform mining directly in most cases because messy redundant content other reasons. This paper analyzes pre-processing on order meet needs mining. At same time, it also puts forward some reasonable processing means.
A massive proliferation of malware variants has posed serious and evolving threats to cybersecurity. Developing intelligent methods cope with the situation is highly necessary due inefficiency traditional methods. In this paper, a efficient, vision-based detection method was proposed. Firstly, bilinear interpolation algorithm utilized for image normalization, data augmentation used resolve issue imbalanced sets. Moreover, paper improved convolutional neural network (CNN) model by combining...
The increasing volume and types of malwares bring a great threat to network security. malware binary detection with deep convolutional neural networks (CNNs) has been proved be an effective method. However, the existing classification methods based on CNNs are unsatisfactory this day because their poor extraction ability, insufficient accuracy classification, high cost time. To solve these problems, novel approach, namely, multiscale feature fusion (MFFCs), was proposed achieve visualization...
In response to escalating cybersecurity threats, this study aims develop a lightweight deep learning model for efficient and accurate detection classification of malicious code. To achieve goal, the introduces TriCh-RepNet, novel network architecture that balances performance resource utilization. The methodology involves innovatively transforming code representations into image channels through three-channel mapping technique, thereby enhancing information richness discriminatory power. By...
With the development of automated malware toolkits, cybersecurity faces evolving threats. Although visualization-based analysis has proven to be an effective method, existing approaches struggle with challenging samples due alterations in texture features binary images during visualization preprocessing stage, resulting poor performance. Furthermore, enhance classification accuracy, methods sacrifice prediction time by designing deeper neural network architectures. This paper proposes PAFE,...
Code debugging is a vital stage of software development, essential for ensuring the reliability and performance Large Language Models (LLMs) in code generation task. Human typically follows multi-stage process, which includes Bug Localization, Identification, Repair, Recognition. However, existing benchmarks predominantly focus on Repair stage, offers only limited perspective evaluating capabilities LLMs. In this paper, we introduce DEBUGEVAL, comprehensive benchmark abilities LLMs by...