A5050526815- Digital Media Forensic Detection
- Advanced Steganography and Watermarking Techniques
- Adversarial Robustness in Machine Learning
- Chaos-based Image/Signal Encryption
- Advanced Image and Video Retrieval Techniques
- Network Security and Intrusion Detection
- Generative Adversarial Networks and Image Synthesis
- Anomaly Detection Techniques and Applications
- Advanced Malware Detection Techniques
- Internet Traffic Analysis and Secure E-voting
- Face recognition and analysis
- Music and Audio Processing
- Video Analysis and Summarization
- Video Coding and Compression Technologies
- Advanced Neural Network Applications
- Biometric Identification and Security
- Privacy-Preserving Technologies in Data
- Software-Defined Networks and 5G
- Advanced Image Processing Techniques
- Handwritten Text Recognition Techniques
- Image Processing Techniques and Applications
- Cybercrime and Law Enforcement Studies
- Advanced Electrical Measurement Techniques
- Opportunistic and Delay-Tolerant Networks
- Video Surveillance and Tracking Methods
Wuhan University
2016-2025
South Central Minzu University
2012
Singapore Management University
2011-2012
Ministry of Education of the People's Republic of China
2009-2011
Nanjing University of Science and Technology
2005-2006
Huazhong University of Science and Technology
2006
The development of smart network infrastructure the Internet Things (IoT) faces immense threat sophisticated Distributed Denial-of-Services (DDoS) security attacks. existing solutions enterprise networks are significantly expensive and unscalable for IoT. integration recently developed Software Defined Networking (SDN) reduces a significant amount computational overhead IoT devices enables additional measurements. At prelude stage SDN-enabled infrastructure, sampling based approach currently...
In this paper, we consider the face swapping detection from perspective of identity. Face aims to replace target with source and generate fake that human cannot distinguish between real fake. We argue contains explicit identity implicit identity, which respectively corresponds during swapping. Note identities faces can be extracted by regular recognizers. Particularly, is consistent its Thus difference facilitates detection. Following idea, propose a novel driven framework for Specifically,...
With the proliferation of smart Internet Things (IoT) devices based on Android system, malicious applications targeting for IoT have received more and attention due to concern privacy leakage property loss. However, existing malware detection approaches static or dynamic analysis are not scalable evolvement cannot extract enough valid semantics in application programming interface (API) level, failing detect new malware. In this paper, we propose EveDroid, a event-aware which exploits...
This paper proposes a framework to improve the existing distortion functions designed for JPEG steganography, which results in better capability of countering steganalysis. Different from steganography approach that minimizes image distortion, we minimize feature caused by data embedding. Given image, construct reference close before compression. Guided both and minimization, state-of-the-art syndrome trellis coding embedding are improved distinguishing costs +1 versus -1 has three...
Abstract Current typical adaptive steganography algorithms take the detection resistant capability into account adequately but usually cannot extract embedded secret messages correctly when stego images suffer from compression attack. In order to solve this problem, a framework of resisting JPEG and is proposed. Utilizing relationship between Discrete Cosine Transformation (DCT) coefficients, domain embedding determined; for maximum ability, modifying magnitude different DCT coefficients...
Current typical adaptive Steganography algorithms cannot extract the embedded secret messages correctly after compression. In order to solve this problem, a JPEG-compression resistant steganography algorithm is proposed. Utilizing relationship between DCT coefficients, domain of embedding determined. The modifying magnitude different coefficients can be determined according quality factors JPEG To ensure completely correct extraction compression, RS codes used encode embedded. Besides, based...
Traditional reversible data hiding (RDH) focuses on enlarging the embedding payloads while minimizing distortion with a criterion of mean square error (MSE). Since imperceptibility can also be achieved via image processing, we propose novel method RDH contrast enhancement (RDH-CE) using histogram shifting. Instead MSE, proposed generates marked images good quality sense structural similarity. The contains two parts: baseline and extensive embedding. In part, first merge least significant...
Recently, it has been discovered that the electric network frequency (ENF) could be captured by digital audio, video, or even image files, and further exploited in forensic investigations. However, existence of ENF multimedia content is not a sure thing, if present, ENF-based analysis would become useless misleading. In this paper, we address problem detection audio recordings, which modeled as weak signal contaminated unknown colored wide-sense stationary (WSS) Gaussian noise, while also...
Deep neural networks are proven to be vulnerable backdoor attacks. Detecting the trigger samples during inference stage, i.e., test-time sample detection, can prevent from being triggered. However, existing detection methods often require defenders have high accessibility victim models, extra clean data, or knowledge about appearance of triggers, limiting their practicality. In this paper, we propose corruption robustness consistency evaluation (TeCo) <sup...
Existing work finds it challenging for adversarial examples to transfer among different synthetic speech detectors because of cross-feature and cross-model. To enhance the transferability examples, we propose a spectral saliency analysis method gain insight into underlying detection mechanisms existing first time. These insights offer an interpretable basis why are between models. Then further two-stage attack framework. Specifically, stage leverages model mechanism design random...
It is well established that neural networks are vulnerable to adversarial examples, which almost imperceptible on human vision and can cause the deep models misbehave. Such phenomenon may lead severely inestimable consequences in safety security critical applications. Existing defenses trend harden robustness of against attacks, for example, training technology. However, these usually intractable implement due high cost retraining cumbersome operations altering model architecture or...
Malicious applications of deep face swapping technology pose security threats such as misinformation dissemination and identity fraud. Some research propose the utilization robust watermarking methods to track copyright facial images, facilitating post-forgery attribution. However, these cannot fundamentally prevent or eliminate adverse impacts swapping. To address this issue, we present Dual Defense, an innovative framework based on adversarial watermarking. It simultaneously tracks image...
Searchable symmetric encryption (SSE) allows the data owner to outsource an encrypted database a remote server in private manner while maintaining ability for selectively search. So far, most existing solutions focus on honest-but-curious server, security designs against malicious have not drawn enough attention. A few recent works attempted construct verifiable SSE that enables verify integrity of search results. Nevertheless, these verification mechanisms are highly dependent specific...