A5101875247- Imbalanced Data Classification Techniques
- Video Surveillance and Tracking Methods
- Advanced Neural Network Applications
- Fire Detection and Safety Systems
- Financial Distress and Bankruptcy Prediction
- Rough Sets and Fuzzy Logic
- Topic Modeling
- Industrial Vision Systems and Defect Detection
- Advanced Steganography and Watermarking Techniques
- Anomaly Detection Techniques and Applications
- Text and Document Classification Technologies
- Machine Fault Diagnosis Techniques
- Advanced Graph Neural Networks
- Image and Signal Denoising Methods
- Advanced Image Fusion Techniques
- Chaos-based Image/Signal Encryption
- Effects of Environmental Stressors on Livestock
- Recommender Systems and Techniques
- Corporate Finance and Governance
- Environmental Quality and Pollution
- Human Pose and Action Recognition
- ECG Monitoring and Analysis
- Domain Adaptation and Few-Shot Learning
- Efficiency Analysis Using DEA
- Animal Nutrition and Physiology
Hefei Institutes of Physical Science
2025
Anhui University
2025
Harbin Engineering University
2014-2025
Shenzhen Technology University
2023-2024
Samsung (United States)
2024
University of Electronic Science and Technology of China
2024
Xidian University
2024
Zhongkai University of Agriculture and Engineering
2020-2023
Northeastern University
2009-2023
University of Michigan
2023
BackgroundMarket-applicable concurrent electrocardiogram (ECG) diagnosis for multiple heart abnormalities that covers a wide range of arrhythmias, with better-than-human accuracy, has not yet been developed. We therefore aimed to engineer deep learning approach the automated multilabel rhythm or conduction by real-time ECG analysis.MethodsWe used dataset ECGs (standard 10 s, 12-channel format) from adult patients (aged ≥18 years), 21 distinct classes, including most types abnormalities,...
Abstract Sparse large-scale multiobjective optimization problems (sparse LSMOPs) are characterized by an enormous number of decision variables, and their Pareto optimal solutions consist a majority variables with zero values. This property sparse LSMOPs presents great challenge in terms how to rapidly precisely search for solutions. To deal this issue, paper proposes adjoint feature-selection-based evolutionary algorithm tailored tackling LSMOPs. The proposed strategy combines two distinct...
In recent years, intelligent fault diagnosis technology with deep learning algorithms has been widely used in industry, and they have achieved gratifying results. Most of these methods require large amount training data. However, actual industrial systems, it is difficult to obtain enough balanced sample data, which pose challenges identification classification. order solve the problems, this paper proposes a data generation strategy based on Wasserstein generative adversarial network...
Modern neural networks can assign high confidence to inputs drawn from outside the training distribution, posing threats models in real-world deployments. While much research attention has been placed on designing new out-of-distribution (OOD) detection methods, precise definition of OOD is often left vagueness and falls short desired notion reality. In this paper, we present a formalization model data shifts by taking into account both invariant environmental (spurious) features. Under such...
Abstract Machine learning methods are widely used to evaluate the risk of small- and medium-sized enterprises (SMEs) in supply chain finance (SCF). However, there may be problems with data scarcity, feature redundancy, poor predictive performance. Additionally, collected over a long time span cause differences distribution, classic supervised exhibit abilities under such conditions. To address these issues, domain-adaptation-based multistage ensemble paradigm (DAMEL) is proposed this study...
Abstract The problem of low-frequency noise is becoming increasingly severe and measuring the sound absorption performance acoustic metamaterials (AMs) using accurate coefficients great interest in control engineering. Conventional calculation methods such as Finite Element Method (FEM) simulations theoretical analysis (TAM) have specific limitations. Deep learning (DL) models provide new perspectives for studying AMs performance. However, prediction DL highly dependent on proper tuning...
Smoke detection based on video surveillance is important for early fire warning. Because the smoke often small and thin in stage of a fire, using collected images identification warning fires very difficult. Therefore, an improved lightweight network that combines attention mechanism upsampling algorithm has been proposed to solve problem stage. Firstly, dataset consists self-created pictures public pictures. Secondly, module combined with channel spatial attention, which are attributes...
Real-time smoke detection is of great significance for early warning fire, which can avoid the serious loss caused by fire. Detecting in actual scenes still a challenging task due to large variance color, texture, and shapes. Moreover, scene faced with difficulties data collection insufficient datasets, morphology susceptible environmental influences. To improve performance solve problem too few datasets real scenes, this paper proposes model that combines deep convolutional generative...
Dianrong, a tech-driven internet finance company, provides loans to large number of people and small business users. The ability predict fraud from loan applications is key the company's business. Based on published literature detection techniques, features have be extracted manually for further rule design or machine learning. But as fraudulent behaviors change over time avoid detection, simple rules become obsolete quickly. Normally, we extract hundreds features, which resource consuming...
Traditional collaborative filtering recommendation algorithms face the cold-start problem. A algorithm based on implicit information of new users and multi-attribute rating matrix is proposed to solve The collected as first-hand interest information. It combined with other create a user-item (UIRM). Singular value decomposition used reduce dimensionality UIRM, resulting in initial neighbor set for target matrix. user ratings are mapped relevant item attributes respectively generate attribute...
Abstract Single image super-resolution (SR) has become a promising research topic, with many deep learning-based models invented to reconstruct high-fidelity high-resolution (HR) images from low-resolution (LR) images. Motivated by large amount of turbulent flow field data collected experimental measurements and numerical simulation, researchers begin investigating the application these data-driven learning conduct SR reconstruction LR data. Due limitations equipment computing power,...
Object detection technology is a crucial research direction in the field of computer vision. Its primary task to identify and locate objects within images. This finds extensive applications across various domains such as autonomous driving, medical diagnostics, security monitoring. With development machine learning, especially progress deep learning related fields image processing, accuracy rate object has achieved better results. However, different algorithms have characteristics, which...
A computer vision technique was used to evaluate internal hollowness, an important textural attribute of Frenchfries. Three algorithms (co-occurrence matrices, texture features corresponding visual perception, andrun-length methods) were modified, implemented, and evaluated on color French fry cross section images. Evaluation andclassification hollowness based in two coordinate systems (RGB I1, I2, I 3).Statistical procedures determine the classification capabilities selected significant...