A5033474275- EEG and Brain-Computer Interfaces
- Blind Source Separation Techniques
- Neural Networks and Applications
- Functional Brain Connectivity Studies
- Human-Automation Interaction and Safety
- Advanced Neural Network Applications
- Advanced Memory and Neural Computing
- Domain Adaptation and Few-Shot Learning
- Epilepsy research and treatment
- Transcranial Magnetic Stimulation Studies
- Multimodal Machine Learning Applications
- Neural and Behavioral Psychology Studies
- Neuroscience and Neural Engineering
- Cancer Treatment and Pharmacology
- Ferroelectric and Negative Capacitance Devices
- Real-time simulation and control systems
- Prostate Cancer Treatment and Research
- Robotic Path Planning Algorithms
- Cancer therapeutics and mechanisms
- Distributed and Parallel Computing Systems
- Heart Rate Variability and Autonomic Control
- Alzheimer's disease research and treatments
- Fault Detection and Control Systems
- Simulation Techniques and Applications
- DNA Repair Mechanisms
China Aerospace Science and Industry Corporation (China)
2021-2024
Sun Yat-sen University
2022
Beihang University
2016-2020
Harvard University
2019-2020
Athinoula A. Martinos Center for Biomedical Imaging
2019-2020
Massachusetts General Hospital
2019-2020
In the automatic detection of epileptic seizures, monitoring critically ill patients with time varying EEG signals is an essential procedure in intensive care units. There increasing interest using analysis to detect seizure, and this study we aim get a better understanding how visualize information time-frequency feature, design train novel random forest algorithm for decoding, especially multiple-levels illness. Here, propose framework seizure based on multiple approaches; it involves...
Automatic recognition methods for non-stationary electroencephalogram (EEG) data collected from EEG sensors play an essential role in neurological detection. The integrated approaches proposed this study consist of Symlet wavelet processing, a gradient boosting machine, and grid search optimizer three-class classification scheme normal subjects, intermittent epilepsy, continuous epilepsy. Fourth-order wavelets are adopted to decompose the into five frequencies sub-bands, such as gamma, beta,...
Abstract Intelligent recognition methods for classifying non-stationary and non-invasive epileptic diagnoses are essential tools in neurological research. Electroencephalogram (EEG) signals exhibit better temporal characteristics the detection of epilepsy compared to radiation medical images like computed tomography (CT) magnetic resonance imaging (MRI), as they provide real-time insights into disease’ condition. While classical machine learning have been used EEG classification, still often...
Abstract Recent studies have suggested that transplant of hiPS-CMs is a promising approach for treating heart failure. However, the optimally clinical benefits been hampered by immature nature hiPS-CMs, and hiPS-CMs-secreted proteins contributing to repair cardiomyocytes remain largely unidentified. Here, we established saponin + compound induced system generate with stable functional attributes in vitro transplanted failure mice. Our study showed enhanced therapeutic effects attenuating...
Abstract The neural basis for long-term behavioral improvements resulting from multi-session transcranial direct current stimulation (tDCS) combined with working memory training (WMT) remains unclear. In this study, we used task-related electroencephalography (EEG) measures to investigate the lasting neurophysiological effects of anodal high-definition (HD)-tDCS applied over left dorsolateral prefrontal cortex (dlPFC) during a challenging WMT. Thirty-four healthy young adults were randomized...
Incremental object detection (IOD) task requires a model to learn continually from newly added data. However, directly fine-tuning well-trained on new will sharply decrease the performance old tasks, which is known as catastrophic forgetting. Knowledge distillation, including feature distillation and response has been proven be an effective way alleviate previous works heavily rely low-level information, while under-exploring importance of high-level semantic information. In this paper, we...
To decode the pilot’s behavioral awareness, an experiment is designed to use aircraft simulator obtaining physiological behavior data. Existing pilot studies such as modeling methods based on domain experts and knowledge discovery do not proceed from characteristics of pilots themselves. The starts directly multimodal explore pilots’ behavior. Electroencephalography, electrocardiogram, eye movement were recorded simultaneously. Extracted features ground missions, air cruise mission trained...
Technology of brain–computer interface (BCI) provides a new way communication and control without language or physical action. Brain signal tracking positioning is the basis BCI research, while brain modeling affects treatment analysis (EEG) functional magnetic resonance imaging (fMRI) directly. This paper proposes human ellipsoid method. Then, we use non-parametric spectral estimation method time–frequency to deal with simulation real EEG epilepsy patients, which utilizes both high spatial...
Automatic recognition methods for non-stationary EEG data collected from sensors play an essential role in neurological detection. The integrative approaches proposed this study consists of Symlet wavelet processing, a gradient boosting machine, and grid search optimizer three-level classification scheme normal subjects, intermittent epilepsy, continuous epilepsy. Fourth-order wavelets were adopted to decompose the into five time-frequency sub-bands, whose statistical features computed used...
This study explored the use of multi-physiological signals and simultaneously recorded high-density electroencephalography (EEG), electrocardiogram (ECG), eye movements to better understand pilots' cognitive behaviour during flight simulator manoeuvres. Multimodal physiological were collected from 12 experienced pilots with international aviation qualifications under wide-angle impressive vision simulation. The data collection spanned two strike missions, each three mission intensities,...
In multi-vehicle systems, the coupling problem between task allocation and path planning variability of execution solutions creates challenges for utility estimation affects effectiveness distributed mission planning. To characterize effect sequences on utilities implement a task-extended allocation, we propose tensor algorithm (TEUTA) based market mechanism. system, consider impact schedule vehicle trajectory, indicate under different preceding points in form tensors. Further, iterative...
A scheduling algorithm is crucial for running a simulation model so that tasks can be performed efficiently.The traditionally used blade-based parallel engine system cannot adapted to new model. This study proposed combined dynamic priority and static method, is, hybrid load-balancing (HLB) algorithm. The given according the operating cycle of steps. experimental results demonstrated outperformed earliest deadline first time-stepped algorithms. also HLBhad higher real-time efficiency than...
Background: Automatic detecting methods for the detection of non-stationary and non-invasive epileptic EEGs are essential tools in neurological research. Traditional machine learning models had been used to classify status epilepsy EEGs, but they still rely on manual features. Meanwhile, earlier some studies focused as a binary (epilepsy vs. healthy subjects) classification rarely ternary (continuous intermittent subjects ) detection.Methods: In this study, we proposed novel deep method,...
Abstract The neural basis for long-term behavioral improvement resulting from multi-session transcranial direct current stimulation (tDCS) combined with working memory training (WMT) remains unclear. In this study, we used task-related electroencephalography (EEG) measures to investigate the lasting neurophysiological effects of anodal high-definition (HD)-tDCS applied over left dorsolateral prefrontal cortex (dlPFC) during a challenging WMT. Thirty-four healthy young adults were randomized...
Model is required to learn from dynamic data stream under incremental object detection task. However, traditional model fails deal with this scenario. Fine-tuning on new task suffers a fast performance decay of early learned tasks, which known as catastrophic forgetting. A promising way alleviate forgetting knowledge distillation, includes feature distillation and response distillation. Previous methods have not discuss selection transfer at the same time. In paper, we propose high-level...