Zhuo Liu

ORCID: 0009-0003-0887-6883
Publications
Citations
Views
---
Saved
---
About
Contact & Profiles
Research Areas
  • Data Stream Mining Techniques
  • Advanced Image and Video Retrieval Techniques
  • Advanced Clustering Algorithms Research
  • Anomaly Detection Techniques and Applications
  • Face and Expression Recognition
  • Extraction and Separation Processes
  • Face recognition and analysis
  • Advanced Battery Technologies Research
  • Image Retrieval and Classification Techniques
  • Imbalanced Data Classification Techniques
  • Advanced Computational Techniques and Applications
  • Recommender Systems and Techniques
  • Adversarial Robustness in Machine Learning
  • Data Management and Algorithms
  • Video Surveillance and Tracking Methods
  • Internet Traffic Analysis and Secure E-voting
  • Medical Image Segmentation Techniques
  • Remote Sensing and Land Use
  • Brain Tumor Detection and Classification
  • Advanced Neural Network Applications
  • Advanced Measurement and Detection Methods
  • Personal Information Management and User Behavior
  • Sparse and Compressive Sensing Techniques
  • Advanced SAR Imaging Techniques
  • Image and Object Detection Techniques

University of Rochester
2024

Xiangtan University
2024

China Mobile (China)
2023-2024

Nanchang University
2020-2023

Xidian University
2021

Guizhou Education University
2020

Chongqing University of Posts and Telecommunications
2020

Dalian Medical University
2020

First Affiliated Hospital of Dalian Medical University
2020

Harbin University of Science and Technology
2016

Acquisition of the standard plane is prerequisite biometric measurement and diagnosis during ultrasound (US) examination. In this paper, a new algorithm developed for automatic recognition fetal facial planes (FFSPs) such as axial, coronal, sagittal planes. Specifically, densely sampled root scale invariant feature transform (RootSIFT) features are extracted then encoded by Fisher vector (FV). The network with multi-layer design also to extract spatial information boost classification...

10.1371/journal.pone.0121838 article EN cc-by PLoS ONE 2015-05-01

Cloud computing is an elastic model that the users can lease resources from rentable infrastructure. gaining popularity due to its lower cost, high reliability and huge availability. To utilize powerful capability of cloud computing, this paper import it into data mining machine learning field. As one most influential open competition in area, Netflix Prize attached with mass storage had driven thousands teams across world attack problem, among which final winner was BellKor's Pragmatic...

10.1109/gcc.2010.89 article EN 2010-11-01

Our objective is to study how obtain features which can reflect the continuity and internal dynamic changes of electroencephalography (EEG) signals an effective method for fatigued driving state recognition based on obtained features.A EEG signalfeature extraction functional data analysis proposed. Combined with kernel principal component method, are applied driver fatigue state, a corresponding model constructed.The tested real collected by selecting suitable classifier. The test results...

10.1088/1361-6579/abc66e article EN Physiological Measurement 2020-10-30

In recent years, the uncertain data stream which is related in many real applications attracts more and attention of researchers. As one aspect character, existence-uncertainty can affect clustering process results significantly. The lately reported algorithms are all based on K-Means algorithm with inhere shortage. DCUStream density-based over proposed this paper. It find arbitrary shaped clusters less time cost high dimension stream. meantime, a dynamic density threshold designed to...

10.1109/fskd.2012.6233800 article EN 2012-05-01

Multi-view feature learning can provide basic information for consistent grouping, and is very common in practical applications, such as judicial document clustering. However, it a challenge to combine multiple heterogeneous features learn comprehensive description of data samples. To solve this problem, many methods explore the correlation between various across views by assuming that all share same semantic information. Inspired this, paper we propose new multi-view robust (MRFL) method....

10.1109/lsp.2020.3026943 article EN IEEE Signal Processing Letters 2020-01-01

This paper describes an automatic recognition approach for facial expression from nearly front view face image. work has two contributions. The first is a method adaptively generating the initial model Active Appearance Models (AAM) fitting, which allows features under large variation to be detected more accuracy. second introduction of set hybrid features, consist geometric AAM shape and appearance parameters. These can optimized when adopting quadratic mutual information (QMI) feature...

10.1016/j.egypro.2012.02.317 article EN Energy Procedia 2012-01-01

10.1109/cvpr52733.2024.02291 article EN 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2024-06-16

The optical network encompasses numerous devices and links, generating a significant volume of logs. Analyzing these logs is for optimization, failure diagnosis, health monitoring. However, the large-scale diverse formats present several challenges, including high cost difficulty manual processing, insufficient semantic understanding in existing analysis methods, strict requirements data security privacy. Generative artificial intelligence (GAI) with powerful language generation capabilities...

10.1364/jocn.527874 article EN Journal of Optical Communications and Networking 2024-10-02

In this paper, two novel methods for SAR image super-resolution (SR) are proposed. The main challenge SR reconstruction is speckle noise. For reason, a algorithm termed the importance sampling cubature Kalman filter (ISCKF) proposed to reconstruct high resolution (HR) from series of low-resolution (LR) images. However, as reconstructed usually embraces residual noise visually, we establish nonlinear low-rank optimization model in order further reduce drastically. Correspondingly, an...

10.1109/tgrs.2023.3305025 article EN IEEE Transactions on Geoscience and Remote Sensing 2023-01-01

Covert faults are characterized by the performance parameters falling within normal range, without any observable abnormalities. These types of pose a significant risk as they present no apparent warning signs potential danger. Therefore, it is crucial to establish an efficient covert fault detection method ensure reliable and stable operation optical networks. Data-driven technology, which reveals internal relations data patterns between historical mining analyzing data, offers new...

10.1364/jocn.502937 article EN Journal of Optical Communications and Networking 2023-10-17

Image change detection is based on the analysis in different time from same area of two or more images, detect feature region information changes over time. Remote sensing image has been widely used such as dynamic monitoring forest resources monitoring, land cover and use, agricultural survey, urban planning layout, environmental analysis, assessment natural disasters, geographic data update military reconnaissance strategic objectives (such roads, Bridges, airports) many other fields....

10.1109/cesys.2016.7889919 article EN 2016-10-01

Randomized weights neural networks have fast learning speed and good generalization performance with one single hidden layer structure. Input weighs of the are produced randomly. By employing certain activation function, outputs calculated some randomization. Output computed using pseudo inverse. Mutual information can be used to measure mutual dependence two variables quantitatively based on probability theory. In this paper, these layer's that relate prediction variable closely selected...

10.4236/jcc.2015.311030 article EN Journal of Computer and Communications 2015-01-01

Adversarial examples, crafted by adding perturbations imperceptible to humans, can deceive neural networks. Recent studies identify the adversarial transferability across various models, \textit{i.e.}, cross-model attack ability of samples. To enhance such transferability, existing input transformation-based methods diversify data with transformation augmentation. However, their effectiveness is limited finite number available transformations. In our study, we introduce a novel approach...

10.48550/arxiv.2405.14077 preprint EN arXiv (Cornell University) 2024-05-22

10.1109/cvci63518.2024.10830067 article EN 2021 5th CAA International Conference on Vehicular Control and Intelligence (CVCI) 2024-10-25

In this paper a method to establish Virtual Chinese Human Eye was presented. method, five datasets were established base on eye tissue slices. 3-D structure model, physical and physiological model of eyeball built up. Someone them can be used in education clinic. The research still awaits long-term will have major effect

10.1109/iembs.2005.1617082 article EN 2005-01-01
Coming Soon ...