A5022948886- Sparse and Compressive Sensing Techniques
- Tensor decomposition and applications
- Fault Detection and Control Systems
- Matrix Theory and Algorithms
- Rough Sets and Fuzzy Logic
- Remote-Sensing Image Classification
- Remote Sensing and Land Use
- Image and Signal Denoising Methods
- Advanced Neuroimaging Techniques and Applications
- Advanced Algorithms and Applications
- Stability and Control of Uncertain Systems
- Advanced Control Systems Optimization
- Transportation Planning and Optimization
- Target Tracking and Data Fusion in Sensor Networks
- Graph theory and applications
- Fluid Dynamics and Turbulent Flows
- Multi-Criteria Decision Making
- Glycosylation and Glycoproteins Research
- Bayesian Modeling and Causal Inference
- Text and Document Classification Technologies
- Network Security and Intrusion Detection
- Reliability and Maintenance Optimization
- Evaluation and Optimization Models
- Advanced Graph Theory Research
- Data Stream Mining Techniques
Fuyang Normal University
2023
Henan University of Science and Technology
2023
UCLouvain
2022
Shandong Normal University
2018-2021
Fuzhou University
2020-2021
China Academy of Space Technology
2021
Newcastle University
2020
National University of Singapore
2018
Northeast Normal University
2010-2017
Weifang People's Hospital
2015
In remote sensing image classification, distance measurements and classification criteria are equally important; less accuracy of either would affect accuracy. Remote was performed by combining support vector machine (SVM) [Formula: see text]-nearest neighbor (KNN). This based on the separability classes using SVM spatial spectral characteristics data. Moreover, a formula is proposed as measure criterion that considers both luminance direction vectors. First, trained vectors (SVs) obtained...
In this article, we study the recursive algorithms for a class of separable nonlinear models (SNLMs) in which parameters can be partitioned into linear part and part. Such are very common machine learning, system identification, signal processing. Utilizing special structure SNLMs, propose variable projection (RVP) algorithm, at each recursion, model eliminated, updated by Levenberg–Marquart algorithm. Then, based on parameters, least-squares According to convergence analysis RVP parameter...
This paper revisits the problem of finding best rank-1 approximation to a symmetric tensor and makes three contributions. First, in contrast many long lingering arguments literature, it offers straightforward justification that generically is symmetric. Second, typical workhorse practice for low-rank approximation, namely, alternating least squares (ALS) technique which improves one factor time, this proposes alternative algorithms based on singular value decomposition (SVD) modifies two...
In this paper we study the orthogonal low-rank approximation problem of tensors in general setting sense that more than one matrix factor is required to be mutually orthonormal, which includes completely and semiorthogonal as two special cases. It has been addressed [L. Wang M. T. Chu, SIAM J. Matrix Anal. Appl., 35 (2014), pp. 1058--1072] "the question semi-orthogonal matrix, except for case complete orthogonality, remains open." To deal with open present an SVD-based algorithm. Our...
Analysis and classification for remote sensing landscape based on imagery is a popular research topic. In this paper, we propose new data classifier by incorporating the support vector machine (SVM) learning information into K-nearest neighbor (KNN) classifier. The SVM well known its extraordinary generalization capability even with limited samples, it very useful applications as samples are usually limited. KNN has been widely used in due to simplicity effectiveness. However, instance-based...
The belief-rule-base (BRB) inference methodology using the evidential reasoning (ER) approach is widely used in different fields, such as fault diagnosis, system identification, and decision analysis. However, calculation characteristic of conventional rule activation weight makes have zero problem. difficulty constructing partial derivatives restricts optimization parameters gradient method. Hence, this paper proposes a new belief structure its training method to solve problem during...
We propose new Riemannian preconditioned algorithms for low-rank tensor completion via the polyadic decomposition of a tensor. These exploit non-Euclidean metric on product space factor matrices in form. This is designed using an approximation diagonal blocks Hessian cost function, thus has preconditioning effect these algorithms. prove that proposed gradient descent algorithm globally converges to stationary point problem, with convergence rate estimates $\L{}$ojasiewicz property. Numerical...
This paper considers the problem of attitude determination automated carrier given only measurements from a low-cost inertial measurement unit (IMU). During systEm modeling and design, quaternion representation is chosen to represent attitude. In ordet correct drift caused by cumulative error gyroscope, we combine gyroscope accelerometer data based on gradient descent complementary filter, which can reduce load MeU improve real timing costing less computation amounts. The experimental...
Implicit feedback techniques take advantage of user behavior to understand interests. The primary use implicit is that such remove the cost providing feedback. So, it cornerstone both search engine and information retrieval. In order better behaviors, this paper investigates three most important issues using assist engines watch any special needs individual user. We have reviewed a number papers presented study results, mainly survey types extensively investigated as sources include...
Purpose: To examine the anticancer effect of salvianolic acid-A against human small cell lung cancer (SCLC) cells (H-69).Methods: In vitro antiproliferative acid SCLC lines was evaluated by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay. Apoptosis induction in these measured videomicroscopy along with fluorescence microscopy using Hoechst 33258 staining. The compound on mitochondrial membrane potential loss detected flow cytometry rhodamine-123 as fluorescent probe....
In article 2203359 Haobo Li, Tinghui Yin, Ziqing Hei, Weifeng Yao, and co-workers, Xe-Pla-MBs is injected targeted to the damaged renal vascular endothelium with platelet-mimicking properties.The destructive pulse of clinical ultrasound system used release inner xenon gas from Xe-Pla-MBs.Xenon enters tubular epithelial cells through temporal intercellular gap endothelial significantly improves function by downregulating cellular senescent signaling.
Positive and unlabeled learning (PU Learning) is a special semi-supervise method. Its most important feature that training set only includes two parts: positive examples examples. Many real-world classification applications appeal to PU Learning problem. The K-means++ clustering algorithm proposed new seeding This paper describes semi-supervised for learning). Our approach extends K-means++, an enhancement K-means seeds the with suitably chosen cluster centers, such situations. experiments...
Integrating quadratic form of algebraic connectivity and Perron value bottleneck matrices, we investigate how the a connected weighted graph behaves under shifting components. Generally speaking, when shift components not containing characteristic vertex from less positive (larger negative) valuation vertices to larger (less vertices, or reduce weights some edges, add new blocks, its is nonincreasing; along paths blocks other block closer (vertex), increase delete non-decreasing. Therefore,...
We consider a Canonical Polyadic (CP) decomposition approach to low-rank tensor completion (LRTC) by incorporating external pairwise similarity relations through graph Laplacian regularization on the CP factor matrices. The usage of entails benefits in learning accuracy LRTC, but at same time, induces coupling terms that hinder optimization model. In order solve graph-regularized we propose efficient alternating minimization algorithms leveraging block structure underlying...
The avalanche effect is an important performance that any block cipher must have. Based on the analysis of KASUMI and with algorithm program experiments, changed ciphertext bits numbers are given by changing every bit plaintext. 64-bit plaintext presented detail test data in this paper.
Considering the characteristics of remote sensing images, incorporating spatial information into traditional pattern recognition technology can improve image classification performance. We utilized statistical region merging (SRM) to obtain similarity between pixels. Then we combined with spectral redefine distance pixels, getting a hybrid measure. To verify effect obtained by SRM on classification, applied and two classifiers based distance: optimum-path forest (OPF) k-nearest neighbours...
The general framework of capacity and travel time reliability urban road networks is proposed by introducing the zonal reserve as a measure index to network service level. Mathematically, formulated bilevel model, in which requirements are converted into explicit link constraints adding dummy links origin-destination (O-D) pairs. integrated well defined under circumstances stochastic variation or demand pattern. model generalizes it also includes special case. Finally, Monte Carlo simulation...
Due to all kinds of need customers and the complicated transmitting environment digital image video resources, numerous practical applications emerge, e.g. Image in painting, interpolation, super-resolution removal salt pepper noise. One thing these cases have common is that there are plenty missing pixels randomly distributed an image. Existing restoration methods aiming at solving this problem include kernel regression [1], matrix completion [2] total variation (TV) model [3]. The 3-D AR...