A5020231974- Fault Detection and Control Systems
- Control Systems and Identification
- Gene expression and cancer classification
- Neural Networks Stability and Synchronization
- Distributed Control Multi-Agent Systems
- Nonlinear Dynamics and Pattern Formation
- Model Reduction and Neural Networks
- Cancer Genomics and Diagnostics
- Structural Health Monitoring Techniques
- COVID-19 diagnosis using AI
- Radiomics and Machine Learning in Medical Imaging
- Neural Networks and Applications
- Advanced Control Systems Optimization
- Bioinformatics and Genomic Networks
- Statistical and numerical algorithms
- Hydraulic and Pneumatic Systems
- Earthquake Detection and Analysis
- Seismic Imaging and Inversion Techniques
- Hydrological Forecasting Using AI
- Machine Fault Diagnosis Techniques
- Advanced Vision and Imaging
- AI in cancer detection
- Brain Tumor Detection and Classification
- Ocean Waves and Remote Sensing
- Lung Cancer Diagnosis and Treatment
Dalian University of Technology
2014-2025
Dalian University
2014-2024
Sichuan University
2016-2018
Chengdu University
2016
Harbin Medical University
2014
First Affiliated Hospital of Harbin Medical University
2014
Renji Hospital
2014
Shanghai Jiao Tong University
2014
Guangdong University of Technology
2014
Kyushu University
2007-2013
In this paper, we propose a decentralized adaptive robust controller for trajectory tracking of robot manipulators. each local controller, disturbance observer (DOB) is introduced to compensate the low-passed coupled uncertainties, and an sliding mode control term employed handle fast-changing components uncertainties beyond pass-band DOB. contrast most controllers using DOB manipulators that are based on linear theory, in study, by some special nonlinear damping terms, boundedness signals...
Radioresistance is the major cause of cancer treatment failure. Additionally, splicing dysregulation plays critical roles in tumorigenesis. However, involvement alternative resistance cells to radiotherapy remains elusive. We sought investigate key role factor SRSF1 radioresistance lung cancer.Lung cell lines, xenograft mice models, and RNA-seq were employed study detailed mechanisms radioresistance. Clinical tumor tissues TCGA dataset utilized determine expression levels distinct...
Anticancer drug responses can be varied for individual patients. This difference is mainly caused by genetic reasons, like mutations and RNA expression. Thus, these features are often used to construct classification models predict the response. research focuses on feature selection issue models. Because of vast dimensions space predicting response, autoencoder network was first built, a subset inputs with important contribution selected. Then using Boruta algorithm, further small set...
Background Pneumothorax can lead to a life-threatening emergency. The experienced radiologists offer precise diagnosis according the chest radiographs. localization of pneumothorax lesions will help quickly diagnose, which be benefit for patients in underdevelopment areas lack radiologists. In recent years, with development large neural network architectures and medical imaging datasets, deep learning methods have become methodology choice analyzing images. objective this study was construct...
Ultrasound (US) imaging is a main modality for breast disease screening. Automatically detecting the lesions in US images essential developing artificial-intelligence-based diagnostic support technologies. However, intrinsic characteristics of ultrasound imaging, like speckle noise and acoustic shadow, always degenerate detection accuracy. In this study, we developed deep learning model called BUSnet to detect tumor with high We first two-stage method including unsupervised region proposal...
Cervical cancer has poor prognosis and patients are often diagnosed at advanced stages of the disease with limited treatment options. There is thus an urgent need for discovery new therapeutic strategies in cervical cancer. The activation SGK1 been linked to development various types but little known about role its potential as a target. Here we report that antioxidative factor promotes survival cells. Gene set enrichment analysis RNA-Seq data reveals strong inverse association between...
Variational mode decomposition (VMD) provides a feasible approach to decompose vibration signals obtained from complex machinery for further applications. The frequency bandwidth control parameter and the total number of modes are critical parameters VMD. Thus, optimally automatically setting two is an essential issue VMD various practical signal sources. To this end, work proposes automatic signals. First, we use evaluation criterion mean mode-location distance evaluate sparsity modes;...
Thrust estimation is a significant part of aeroengine thrust control systems. The traditional methods are either low in accuracy or large computation. To further improve the effect, estimator based on Multi-layer Residual Temporal Convolutional Network (M−RTCN) proposed. solve problem dead Rectified Linear Unit (ReLU), proposed method uses Gaussian Error (GELU) activation function instead ReLU residual block. Then overall architecture multi-layer convolutional network adjusted by using...
AbstractIn this paper, we propose a distributed robust control method for synchronised tracking of networked Euler–Lagrange systems, where the time-varying reference trajectory is sent to only subset agents. It assumed that agents can exchange information with their local neighbours on bidirectionally connected communication graph. In controller equipped in each generalised coordinate agents, disturbance observer introduced compensate low-passed-coupled uncertainties, and sliding mode term...
Identification of driver genes, whose mutations cause the development tumors, is crucial for improvement cancer research and precision medicine. To overcome problem that traditional frequency-based methods cannot detect lowly recurrently mutated researchers have focused on functional impact gene proposed function-based methods. However, most estimate distribution null model through non-parametric method, which sensitive to sample size. Besides, such could probably lead underselection or...
Predicting significant wave height (SWH) is for coastal energy evaluation and utilization, port construction, shipping planning. It has been reported that SWH difficult to forecast the complex marine conditions chaos in nature. Current methods either require reliable prior information or reach upper limit of prediction accuracy. To this end, paper proposes a wavelet-based residual network predict with high First, time-series data wave-related factors collected by ocean buoy station...
We address a novel predictive control strategy for dual-rate systems in which the input updating period is different from output sampling based on lifted state-space model identified by modified Numerical Subspace State-Space IDentification (N4SID). There are three steps strategy. Firstly, models N4SID. Based model, we construct predictors can predict of multi-step ahead. Combining with an objective function minimization, laws derived.
Due to the heterogeneity of cancer, precision medicine has been a major challenge for cancer treatment. Determining medication regimens based on patient genotypes become research hotspot in genomics. In this study, we aim identify key biomarkers targeted therapies single nucleotide variants (SNVs) and copy number (CNVs) genes. The experiment is carried out 7 cancers Encyclopedia Cancer Cell Lines (CCLE) dataset. Considering high mutability driver genes which result abundant mutated samples,...
Abstract The combination of support vector machine (SVM) and deep learning is widely used in bearing fault diagnosis. SVM-based diagnosis method usually uses features extracted from vibration signals as input. However, environmental noise industrial applications can affect the feature representation, potentially leading to failures methods. Adversarial robustness verification methods evaluate models small perturbations representations. Certified defense achieve enhancement by embedding...
Video semantic segmentation is a challenging vision task since the temporal-spatial characteristics are difficult to model satisfy requirements of real-time and accuracy simultaneously. To tackle this problem, paper proposes novel optical flow based method. We propose an adaptive threshold key frame scheduling strategy temporal information by estimating inter-frame similarity. ensure accuracy, we construct convolutional neural network named Quick Network with attention (QNet-attention),...
In the pursuit of precision medicine for cancer, a promising step is to predict drug response based on data mining, which can provide clinical decision support cancer patients. Although some machine learning methods predicting from genomic already exist, most them focus point prediction, cannot reveal distribution predicted results. this paper, we propose three-layer feature selection combined with gamma GLM and two-layer an ANN. The two regression are applied Encyclopedia Cancer Cell Lines...
The disturbance storm time (Dst) index is a measure of the geomagnetic strength that can be caused by solar wind plasma ejecta and/or high-speed streams. research aims to predict Dst hours ahead using statistical regression models based on measurements. It shown distribution data has heavy tails. This implies cannot well approximated with Gaussian distribution. Instead, we use t-distributions model data. By considering Sun-earth coupling process as stochastic dynamical system, construct...