A5100323010- Imbalanced Data Classification Techniques
- Geochemistry and Geologic Mapping
- Nuclear Physics and Applications
- Financial Distress and Bankruptcy Prediction
- Metabolomics and Mass Spectrometry Studies
- Atmospheric Ozone and Climate
- Digital Imaging for Blood Diseases
- Vehicle License Plate Recognition
- Topic Modeling
- Ideological and Political Education
- Insurance and Financial Risk Management
- Electricity Theft Detection Techniques
- Algorithms and Data Compression
- Anomaly Detection Techniques and Applications
- Medical Research and Treatments
- Advanced Technologies in Various Fields
- Explainable Artificial Intelligence (XAI)
- Digital Media Forensic Detection
- Laser-Ablation Synthesis of Nanoparticles
- Calibration and Measurement Techniques
- Engineering and Technology Innovations
- Educational Robotics and Engineering
- Cognitive Science and Mapping
- Orbital Angular Momentum in Optics
- Machine Learning in Bioinformatics
Nanjing University
2010-2025
Guilin University of Technology
2022-2023
Dalian University of Technology
2023
Kennesaw State University
2019
Guangzhou Marine Geological Survey
2014
Max Planck Institute for Gravitational Physics
2010
Georgetown University
2004
Class-Incremental Learning (CIL) or continual learning is a desired capability in the real world, which requires system to adapt new tasks without forgetting former ones. While traditional CIL methods focus on visual information grasp core features, recent advances Vision-Language Models (VLM) have shown promising capabilities generalizable representations with aid of textual information. However, when continually trained classes, VLMs often suffer from catastrophic knowledge. Applying poses...
Abstract During the past decade, metasurfaces have provided a versatile platform for controlling amplitude, phase, and polarization of electromagnetic waves, thus showing great potential developing compact photonic devices. However, broadband terahertz (THz) components are still scarce high‐speed communication spectroscopy analysis. Here, free‐standing metasurface is designed to realize efficient wavefront manipulation in THz regime. The proposed consists single patterned metallic layer on...
Gene selection, cancer classification and functional gene are three main concerns interests by biologists for detection, classification, understanding the functions of genes from molecular level tissues, where large number relatively small experiments in expression data generate a great challenge. After brief introduction support vector machine(SVM) this paper presents recent SVM approaches followed analysis on advantages limitations these applications.
Abstract Civic and political education work in colleges universities is a reform measure of human put forward under the fundamental pursuit urgent demand for by development country society. The article first researches relevant theories factor analysis method, completes construction step-by-step design mathematical model evaluation index Political Education, further constructs CIPP model. Finally, it takes Education School H as an empirical study. It tries to carry out panoramic so improve...
Predictive Autoscaling is used to forecast the workloads of servers and prepare resources in advance ensure service level objectives (SLOs) dynamic cloud environments. However, practice, its prediction task often suffers from performance degradation under abnormal traffics caused by external events (such as sales promotional activities applications re-configurations), for which a common solution re-train model with data long historical period, but at expense high computational storage costs....
We aim at developing and improving the imbalanced business risk modeling via jointly using proper evaluation criteria, resampling, cross-validation, classifier regularization, ensembling techniques. Area Under Receiver Operating Characteristic Curve (AUC of ROC) is used for model comparison based on 10-fold cross validation. Two undersampling strategies including random (RUS) cluster centroid (CCUS), as well two oversampling methods (ROS) Synthetic Minority Oversampling Technique (SMOTE),...
Large Language Models (LLMs) have shown impressive performance in natural language tasks, but their outputs can exhibit undesirable attributes or biases. Existing methods for steering LLMs toward desired often assume unbiased representations and rely solely on prompts. However, the learned from pre-training introduce semantic biases that influence process, leading to suboptimal results. We propose LLMGuardrail, a novel framework incorporates causal analysis adversarial learning obtain LLMs....
We aim at developing and improving the imbalanced business risk modeling via jointly using proper evaluation criteria, resampling, cross-validation, classifier regularization, ensembling techniques. Area Under Receiver Operating Characteristic Curve (AUC of ROC) is used for model comparison based on 10-fold cross validation. Two undersampling strategies including random (RUS) cluster centroid (CCUS), as well two oversampling methods (ROS) Synthetic Minority Oversampling Technique (SMOTE),...