A5106976370- Hand Gesture Recognition Systems
- Anomaly Detection Techniques and Applications
- Gait Recognition and Analysis
- Muscle activation and electromyography studies
- Medical Image Segmentation Techniques
- Image and Object Detection Techniques
- Optical measurement and interference techniques
- Human Pose and Action Recognition
- Industrial Vision Systems and Defect Detection
- Advanced Steganography and Watermarking Techniques
- Dental Radiography and Imaging
- Advanced Image and Video Retrieval Techniques
- Biomedical Text Mining and Ontologies
- Risk and Safety Analysis
- IoT and Edge/Fog Computing
- Domain Adaptation and Few-Shot Learning
- Evaluation Methods in Various Fields
- Digital Media Forensic Detection
- Privacy-Preserving Technologies in Data
- Education and Learning Interventions
- Authorship Attribution and Profiling
- Gaussian Processes and Bayesian Inference
- Advanced Neural Network Applications
- Advanced Decision-Making Techniques
- Hearing Impairment and Communication
Fudan University
2023-2025
Beijing University of Posts and Telecommunications
2023-2024
Chiang Mai University
2020-2024
State Key Laboratory of Networking and Switching Technology
2023
Beijing Institute of Technology
2021-2022
Arizona State University
2021-2022
Flexible wearable sensor electronics, combined with advanced software functions, pave the way toward increasingly intelligent healthcare devices. One important application area is limb prosthesis, where printed flexible solutions enable efficient monitoring and assessing of actual intra-socket dynamic operation conditions in clinical other more natural environments. However, data collected by such sensors suffer from variations errors, leading to difficulty perceiving operational conditions....
Sign language, serving as a crucial means of communication for individuals with hearing impairments, plays pivotal role in achieving barrier-free the broader population. Despite advancements utilizing deep learning automatic recognition dynamic sign challenges persist, especially context video-based language recognition. This requires careful consideration spatial and temporal features, along inherent long-term characteristics sequence. In this paper, we introduce novel model framework that...
Sign language is one of the most effective ways to help hearing-impaired people communicate with other people. Although deep learning methods have been used in recognition, there are still problems finger sign recognition. The major issue that gradient approach usually fails, or obtained recognition accuracy not high when depth increasing. We thus propose a Chinese method based on ResNet and Adam optimizer together additional image processing techniques gain higher accuracy. then compare our...
Machine learning research on medical images is becoming popular as advanced imaging technologies and equipment in medicine become more available. Dental Cone-beam Computed Tomography (Dental CBCT), a frequently-used visualization tool for oral diagnosis, provides valuable three-dimensional information, whose development automation of CBCT analysis, the other hand, relatively preliminary. Generally, there are three important characteristics analyzing with noisy labels limited labeled sample...
Sign language recognition technology can help people with hearing impairments to communicate those who are impaired. At present, the rapid development of society, deep learning also provided certain technical support for sign work. In tasks, use traditional convolutional neural networks extract spatio-temporal features from videos suffers insufficient feature extraction, resulting in low rates. Nevertheless, video-based datasets very large and require a lot computational resources training...
The failure rate assessment of online metering equipment is significan t for power metering. For traditional methods, the performance model not satisfactory especially in case small samples. In this paper, a n measuring fault evaluation method based on Weibull parameter hierarchical Bayesian proposed. Firstly, z-score used to eliminate outliers raw data. Then, generalized linear function with variable intercept established according characteristics information each region merged using...
This paper presents a reconfigurable near-sensor anomaly detection processor to real-time monitor the potential anomalous behaviors of amputees with limb prostheses. The is low-power, low-latency, and suitable for equipment on prostheses comprises Variational Autoencoder (VAE), scalable Self-Organizing Map (SOM) Array, window-size-adjustable Markov Chain, which can implement an integrated miniaturized system. With VAE, proposed support up 64 sensor sampling channels programmable by global...
One important characteristic of modern fault classification systems is the ability to flag system when faced with previously unseen types. This work considers unknown detection capabilities deep neural network-based classifiers. Specifically, we propose a methodology on how, available, labels regarding taxonomy can be used increase performance without sacrificing model performance. To achieve this, utilize soft label techniques improve state-of-the-art novel during training process and...
Transfer learning has become an essential technique to exploit information from the source domain boost performance of target task. Despite prevalence in high-dimensional data, heterogeneity and heavy tails are insufficiently accounted for by current transfer approaches thus may undermine resulting performance. We propose a procedure framework quantile regression models accommodate domains. establish error bounds estimator based on delicately selected transferable domains, showing that lower...
6G is envisioned to serve many compute-intensive and latency-sensitive applications which are now usually formed by functional components in the directed acyclic graphic (DAG) form. Mobile edge computing (MEC) proven a promising way offload tasks reduce execution latency. However, ever-increasing number heterogeneity of nodes, as well dynamically changing resource state, hinder any centralized scheduler from getting accurate timely state huge nodes. In this paper, facing large-scale MEC...
The explosive growth of medical images brings about massive amounts high-dimensional images. image retrieval technique is effective in selecting some similar for analysis. A vast majority existing methods aim to extract a single feature represent image. Although these have improved the performance, they ignore that different characteristics express information and are an indispensable part content information. Moreover, conventional features consume much storage, which not conducive data...