- Digital Media Forensic Detection
- Advanced Steganography and Watermarking Techniques
- Antenna Design and Analysis
- Microwave Engineering and Waveguides
- Image Processing Techniques and Applications
- Advanced Image and Video Retrieval Techniques
- Target Tracking and Data Fusion in Sensor Networks
- Image and Object Detection Techniques
- Advanced Measurement and Detection Methods
- Image Retrieval and Classification Techniques
- Radar Systems and Signal Processing
- Face and Expression Recognition
- Face recognition and analysis
- Antenna Design and Optimization
- Advanced SAR Imaging Techniques
- Direction-of-Arrival Estimation Techniques
- Image Processing and 3D Reconstruction
- Advanced Wireless Communication Techniques
- Advanced Image Processing Techniques
- Natural Language Processing Techniques
- Video Surveillance and Tracking Methods
- Advanced Image Fusion Techniques
- Generative Adversarial Networks and Image Synthesis
- Handwritten Text Recognition Techniques
- Piezoelectric Actuators and Control
Wuhan Institute of Technology
2023-2025
Huaiyin Institute of Technology
2024
North China University of Science and Technology
2024
Nanjing University of Science and Technology
2012-2023
National University of Singapore
2023
Nanjing University of Information Science and Technology
2019-2023
Huazhong University of Science and Technology
2019-2023
University of Science and Technology of China
2022
Zhejiang Lab
2021-2022
University of Electronic Science and Technology of China
2020-2021
Collaborative filtering (CF) is a successful approach commonly used by many recommender systems. Conventional CF-based methods use the ratings given to items users as sole source of information for learning make recommendation. However, are often very sparse in applications, causing degrade significantly their recommendation performance. To address this sparsity problem, auxiliary such item content may be utilized. topic regression (CTR) an appealing recent method taking which tightly...
Abstract Transition metal‐based oxyhydroxides (MOOH) derived from the irreversible structural reconstruction of precatalysts are often acknowledged as real catalytic species for oxygen evolution reaction (OER). Typically, reconstruction‐derived MOOH would exhibit superior OER activity compared to their directly synthesized counterparts, despite being fundamentally similar in chemistry. As such, has emerged a promising strategy boost electrocatalysts. However, in‐depth understanding origin...
Despite the recent progress in Graph Neural Networks (GNNs), it remains challenging to explain predictions made by GNNs. Existing explanation methods mainly focus on post-hoc explanations where another explanatory model is employed provide for a trained GNN. The fact that fail reveal original reasoning process of GNNs raises need building with built-in interpretability. In this work, we propose Prototype Network (ProtGNN), which combines prototype learning and provides new perspective...
The unique atom configuration in high-entropy dielectric ceramics gives rise to high P max , small r . Accordingly, ultrahigh recoverable energy density of 8.8 J cm −3 and η 92.5%, as well excellent thermal stability, are achieved.
Detection of double Joint Photographic Experts Group (JPEG) compression is an important part image forensics. Although methods in the past studies have been presented for detecting JPEG with a different quantization matrix, detection same matrix still challenging problem. In this paper, effective method to detect recompression color images by using conversion error, rounding and truncation error on pixel spherical coordinate system proposed. The randomness errors, errors result random...
Nowadays, users upload numerous photos to social network platforms share their daily lives. These contain personal information, which can be easily captured by intelligent algorithms. To improve privacy security, we aim form a protection mechanism exploiting adversarial examples, mislead and disrupt However, the existing attack lacks study on recoverability reversibility, makes them unable serve as an effective mechanism. address this issue, propose recoverable generative generate...
Diffusion models have made significant advancements in recent years. However, their performance often deteriorates when trained or fine-tuned on imbalanced datasets. This degradation is largely due to the disproportionate representation of majority and minority data image-text pairs. In this paper, we propose a general fine-tuning approach, dubbed PoGDiff, address challenge. Rather than directly minimizing KL divergence between predicted ground-truth distributions, PoGDiff replaces...
Image‐to‐image ( I2I ) translation has emerged as a valuable tool for privacy protection in the digital age, offering effective ways to safeguard portrait rights cyberspace. In addition, is applied real‐world tasks such image synthesis, super‐resolution, virtual fitting, and live streaming. Traditional models demonstrate strong performance when handling similar datasets. However, domain distance between two datasets large, quality may degrade significantly due notable differences shape...
Joint photographic experts group (JPEG) compression is widely used in image processing and computer vision. Detecting double compressed JPEG images a common problem forensics detecting with the same quantization matrix remains challenging task. However, most existing methods were designed for detection grayscale cannot fully use unique characteristics of color (such as relationship between channels information). In addition, performance unsatisfactory low quality factors cross experiments....
Classic Preisach model can precisely describe the hysteresis of piezoelectric stack actuators, but its identification is relatively complicated. Neural network easy to be identified with available training algorithm, it cannot directly multi-valued mapping hysteresis. A neural-Preisach was proposed for modeling and control actuators. The inherits advantages neural network, which update parameters by algorithm. feedforward controller designed inverse model, then experiments tracking were...
The quantization step is a crucial parameter in JPEG compression, that can reveal the compression history of image. Estimating steps for single compressed and recompressed images attracting considerable interest field image forensics steganalysis. Several effective methods have been proposed, but performance these still needs to be improved on small-sized low-quality images. To solve above problems, feature enrichment performed frequency domain, resulting clustering discrete cosine transform...
Detection of aligned double Joint Photographic Experts Group (JPEG) compressed images is a crucial area research within the field digital image forensics. The detection tasks for JPEG compression can be categorized into two sub-tasks, namely detecting with same quantization matrix (DJSQM) or different matrices (DJDQM). Existing methods one these sub-tasks may not effective other. To address this issue, novel approach proposed by recompressing both DJDQM and DJSQM using modified coefficients....
This paper studies the problem of transfer learning in context reinforcement learning. We propose a novel method that can speed up with aid previously learnt tasks. Before performing extensive episodes, our attempts to analyze task via some exploration environment, and then reuse previous experience whenever it is possible appropriate. In particular, proposed consists four stages: 1) subgoal discovery, 2) option construction, 3) similarity searching, 4) reusing. Especially, order fulfill...
Presents a reply to comments on the paper, "Optimization of sparse frequency diverse array with time-invariant spatialfocusing beampattern," (Yang, Y.-Q., et al), IEEE Antennas Wireless Propag. Lett., vol. 17, no. 2, pp. 351–354, Feb. 2018.
Early detection and accurate classification of lung nodules are crucial for the treatment cancer.With widespread application deep learning technologies in medical imaging analysis, significant progress has been made automatic from computed tomography (CT) images.However, existing approaches often face challenges with limited annotated data generalization across diverse datasets.To address these challenges, this study introduces two innovative methods: a domain-adaptive adversarial network...
This paper describes a connected component (CC)-based approach to automatic text location and segmentation in natural scene images. A multi-group decomposition scheme is used deal with the complexity of color background. Connected extraction implemented using block adjacency graph (BAG) algorithm after noise filtering runlength smearing (RLS) operation. Some heuristic features priority adaptive (PAS) characters are proposed candidate verification grayscale-based recognition. prototype system...