A5102807602- Face recognition and analysis
- Advanced Adaptive Filtering Techniques
- Biometric Identification and Security
- Speech and Audio Processing
- Control Systems and Identification
- Aeroelasticity and Vibration Control
- Advanced Image and Video Retrieval Techniques
- Blockchain Technology Applications and Security
- Advanced Sensor and Control Systems
- Domain Adaptation and Few-Shot Learning
- Cryptography and Data Security
- Forensic and Genetic Research
- Spectroscopy Techniques in Biomedical and Chemical Research
- Anomaly Detection Techniques and Applications
- User Authentication and Security Systems
- Cloud Data Security Solutions
- Security and Verification in Computing
- Face and Expression Recognition
- Machine Learning and ELM
- Video Surveillance and Tracking Methods
- Multimodal Machine Learning Applications
- Advanced Neural Network Applications
Nanyang Technological University
2025
Peking University
2019-2021
Joint iris-periocular recognition based on feature fusion can overcome some inherent drawbacks of unimodal biometrics, but most the prior works are limited by conventional extraction approaches and fixed schemes. To achieve more accurate adaptive recognition, an end-to-end deep network for joint is proposed in this paper. Multiple attention mechanisms including self-attention co-attention integrated into network. Specifically, two forms mechanisms, spatial channel attention, inserted module,...
The need for data trading promotes the emergence of market. However, in conventional markets, both buyers and sellers have to use a centralized platform which might be dishonest. A dishonest may steal resell seller's data, or refuse send after receiving payment from buyer. It seriously affects fair transaction harm interests parties transaction. To address this issue, we propose novel blockchain-based framework with Trusted Execution Environment (TEE) provide trusted decentralized trading....
Joint face-iris identification can integrate complementary information from face and iris to fulfill the requirement of performance improvement security. However, most current multimodal biometric systems acquire with different sensors which brings about increase capturing complexity device cost. Besides, they are limited by degradation under non-ideal scenarios. In order address these problems, a robust single-sensor system based on feature extraction (MFE) network is proposed. Only single...
Virtual sensing (VS) technology enables active noise control (ANC) systems to attenuate at virtual locations distant from the physical error microphones. Appropriate auxiliary filters (AF) can significantly enhance effectiveness of VS approaches. The selection appropriate AF for various types be automatically achieved using convolutional neural networks (CNNs). However, training CNN model different ANC is often labour-intensive and time-consuming. To tackle this problem, we propose a novel...
The Kalman filter (KF)-based active noise control (ANC) system demonstrates superior tracking and faster convergence compared to the least mean square (LMS) method, particularly in dynamic cancellation scenarios. However, environments with extremely high levels, power of signal can exceed system's rated output due hardware limitations, leading saturation subsequent non-linearity. To mitigate this issue, a modified KF an constraint is proposed. In approach, disturbance treated as measurement...
The need for data trading promotes the emergence of market. However, in conventional markets, both buyers and sellers have to use a centralized platform which might be dishonest. A dishonest may steal resell seller's data, or refuse send after receiving payment from buyer. It seriously affects fair transaction harm interests parties transaction. To address this issue, we propose novel blockchain-based framework with Trusted Execution Environment (TEE) provide trusted decentralized trading....
Recent approaches for real-time semantic segmentation usually employ the encoder-decoder architecture as backbone to generate a high-quality prediction. There has been lot of research on designing efficient encoding methods. However, enhancing performance components in decoder is also crucial pixel-level recognition. In this paper, we propose self-learned feature reconstruction (SFR) method and an offset-dilated fusion (ODFF) module improve prediction capability decoder. Concretely, SFR can...
Hashing methods have been widely applied to approximate nearest neighbor search for large-scale image retrieval, due its computation efficiency and retrieval quality. Deep hashing can improve the quality by representation learning hash coding. Existing deep only take spatial features into account result in lack of accurate semantic similarities images pairs. In this paper, a novel network, Captioning Network (DCHN), is proposed enhance codes. DCHN, binary codes are generated Bayesian...