A5072912487- Advanced Image Processing Techniques
- Rough Sets and Fuzzy Logic
- Image Enhancement Techniques
- Advanced Neural Network Applications
- Data Mining Algorithms and Applications
- Domain Adaptation and Few-Shot Learning
- Advanced Image Fusion Techniques
- Image and Signal Denoising Methods
- Advanced Vision and Imaging
- Advanced Image and Video Retrieval Techniques
- Generative Adversarial Networks and Image Synthesis
- Advanced Algebra and Logic
- Adversarial Robustness in Machine Learning
- Multi-Criteria Decision Making
- Data Management and Algorithms
- Advanced Computational Techniques and Applications
- Multimodal Machine Learning Applications
- Video Surveillance and Tracking Methods
- Human Pose and Action Recognition
- Anomaly Detection Techniques and Applications
- Digital Media Forensic Detection
- Remote-Sensing Image Classification
- AI in cancer detection
- Image Processing Techniques and Applications
- Visual Attention and Saliency Detection
Institute of Software
2023-2025
China University of Petroleum, East China
2016-2025
China University of Petroleum, Beijing
2021
Qingdao University of Technology
2011-2014
Shihezi University
2010-2014
Nanjing Medical University
2014
Jiangxi University of Finance and Economics
2006-2007
Tsinghua University
2006-2007
Xi'an Jiaotong University
2004-2005
A fuzzy rough set is an important model used for feature selection. It uses the dependency as a criterion However, this can merely maintain maximal function. does not fit given dataset well and cannot ideally describe differences in sample classification. Therefore, study, we introduce new handling problem. First, define decision of using concept neighborhood. Then, parameterized relation introduced to characterize information granules, which lower upper approximations are reconstructed...
The concept of dependency in a neighborhood rough set model is an important evaluation function for the feature selection. This considers only classification information contained lower approximation decision while ignoring upper approximation. In this paper, we construct class uncertainty measures: self-information These measures take into account and approximations. relationships between these their properties are discussed detail. It proven that fourth measure, called relative...
Classical rough set theory is considered a useful tool for dealing with the uncertainty of categorical data. The major deficiency this model that classical sensitive to noise in classification learning due stringent condition equivalence relation. Thus, class fuzzy similarity relations was introduced describe between samples attributes. However, these kinds also have deficiencies when they are used computation. In article, we propose new fuzzy-rough-set data by introducing variable parameter...
Since the training data of target model is not available in black-box substitute attack, most recent schemes utilize generative adversarial networks (GANs) to generate for model. However, these GANs-based suffer from low efficiency as generator needs be retrained each during process, well generation quality. To overcome limitations, we consider utilizing diffusion (DM) and propose a novel data-free attack scheme based on stable (SD) improve accuracy training. Despite generated by SD...
Rough sets theory has proved to be a useful mathematical tool for classification and prediction. However, as many real-world problems deal with ordering objects instead of classifying objects, one the extensions classical rough approach is dominance-based approach, which mainly based on substitution indiscernibility relation by dominance relation. In this article, we present reasoning in incomplete ordered information systems. The shows how find decision rules directly from an table. We...
In this paper, we introduce the notion of formal decision context as an extension contexts by employing information table. We use concept analysis to formulate approach extract "if–then" rule from contexts. also construct a knowledge-lossless method for complexity reduction in so that maximum rules extracted reduced are identical initial More specifically, develop discernibility matrix and function compute all attribute reductions without loss knowledge.
The Adaptive Boosting (AdaBoost) algorithm is a widely used ensemble learning framework, and it can get good classification results on general datasets. However, challenging to apply the AdaBoost directly imbalanced data since designed mainly for processing misclassified samples rather than of minority classes. To better process data, this paper introduces indicator Area Under Curve (AUC) which reflect comprehensive performance model, proposes an improved based AUC (AdaBoost-A) improves...
Raindrops adhered to a glass window or camera lens appear in various blurring degrees and resolutions due the difference of raindrops aggregation. The removal from rainy image remains challenging task because density diversity raindrops. abundant location blur level information are strong prior guide raindrop removal. However, existing methods use binary mask locate estimate with value 1 (adhesion raindrops) 0 (no adhesion), which ignores Meanwhile, it is noticed that different scale...
The field of remote sensing (RS) image change detection (CD) has made significant progress, largely due to the powerful feature representation abilities deep learning. However, traditional methods have not fully exploited valuable information in differences. These often treat models as tools extract features from individual images, which limits their ability effectively describe Additionally, many approaches tend focus on spatial differences, while neglecting variations channel dimension. In...