A5043525038- 3D Shape Modeling and Analysis
- Sleep and Work-Related Fatigue
- Genetic Associations and Epidemiology
- Network Security and Intrusion Detection
- Computer Graphics and Visualization Techniques
- Domain Adaptation and Few-Shot Learning
- Brain Tumor Detection and Classification
- Cancer-related molecular mechanisms research
- Non-Invasive Vital Sign Monitoring
- Imbalanced Data Classification Techniques
- Advanced Neural Network Applications
- RNA Research and Splicing
- Obstructive Sleep Apnea Research
- Advanced Image and Video Retrieval Techniques
- 3D Surveying and Cultural Heritage
- Anomaly Detection Techniques and Applications
Ningxia University
2022-2024
Guangzhou Medical University
2024
Abstract Objective . Snoring is the most typical symptom of obstructive sleep apnea hypopnea syndrome (OSAHS) that can be used to develop a non-invasive approach for automatically detecting OSAHS patients. Approach In this work, model based on transfer learning and fusion was applied classify simple snorers Three kinds basic models were constructed pretrained Visual Geometry Group-16 (VGG16), audio neural networks (PANN), Mel-frequency cepstral coefficient (MFCC). The XGBoost select features...
Genome-wide association studies (GWAS) have identified thousands of variants in the human genome with autoimmune diseases. However, identifying functional regulatory associated diseases remains challenging, largely because insufficient experimental validation data. We adopt concept semi-supervised learning by combining labeled and unlabeled data to develop a deep learning-based algorithm framework, sscNOVA, predict analyze characteristics these variants. Compared traditional supervised...
A common study area in anomaly identification is industrial images detection based on texture background. The interference of and the minuteness anomalies are main reasons why many existing models fail to detect anomalies. We propose a strategy for that combines dictionary learning normalizing flow aforementioned questions. two-stage approach already use enhanced by our method. In order improve baseline method, this research adds representation deep learning. Improved algorithms have...
A 3D reconstruction method based on dynamic graph convolutional occupancy networks is proposed to address the issues of texture information loss, geometric loss after voxelization, and lack object completeness constraints in process using voxel representation a block-wise manner. By constructing structure for feature extraction, aims at restore models with fewer holes local details. In extraction stage, pooling employed within each point cloud block problem nonsignificant loss. To tackle...
Structural network pruning is an effective way to reduce size for deploying deep networks resource-constrained devices. Existing methods mainly employ knowledge distillation from the last layer of guide whole network, and informative features intermediate layers are not yet fully exploited improve efficiency accuracy. In this paper, we propose a block-wisely supervised (BNP) approach find optimal subnet baseline based on Markov Chain Monte Carlo. To achieve this, divided into small blocks,...