检索结果-南通市图书馆

A Deep Learning Hierarchical Ensemble for Remote Sensing Image Classification

维普期刊数据库

同方期刊数据库评论

在线全文

学校读者我要写书评

暂无评论

Computers, Materials & Continua 2022年第8期72卷 2649-2663页

作者： Seung-Yeon Hwang jeong-Joon Kim Department of Computer Engineering Anyang UniversityAnyang-si14028Korea Department of ICT Convergence Engineering Anyang UniversityAnyang-si14028Korea

Artificial intelligence,which has recently emerged with the rapid development of information technology,is drawing attention as a tool for solving various problems demanded by society and *** particular,convolutional neural networks(CNNs),a type of deep learning technology,are highlighted in computer vision fields,such as image classification and recognition and object *** these CNN models requires a large amount of data,and a lack of data can lead to performance degradation problems due to *** CNN architecture development and optimization studies become active,ensemble techniques have emerged to perform image classification by combining features extracted from multiple CNN *** this study,data augmentation and contour image extraction were performed to overcome the data shortage *** addition,we propose a hierarchical ensemble technique to achieve high image classification accuracy,even if trained from a small amount of ***,we trained the UCMerced land use dataset and the contour images for each image on pretrained VGGNet,GoogLeNet,ResNet,DenseNet,and *** then apply a hierarchical ensemble technique to the number of cases in which each model can be *** experiments were performed in cases where the proportion of training datasets was 30%,50%,and 70%,resulting in a performance improvement of up to 4.68%compared to the average accuracy of the entire model.

关键词： Image classification deep learning CNNs hierarchical ensemble UC-Merced land use dataset contour image

The Impact of Semi-Supervised Learning on the Performance of Intelligent Chatbot System

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Computers, Materials & Continua 2022年第5期71卷 3937-3952页

作者： Sudan Prasad Uprety Seung Ryul jeong The Graduate School of Business Information Technology Kookmin UniversitySeoul02707Korea

Artificial intelligent based dialog systems are getting attention from both business and academic *** key parts for such intelligent chatbot systems are domain classification,intent detection,and named entity *** supervised,unsupervised,and hybrid approaches are used to detect each *** intelligent systems,also called natural language understanding systems analyze user requests in sequential order:domain classification,intent,and entity recognition based on the semantic rules of the classified *** sequential approach propagates the downstream error;i.e.,if the domain classification model fails to classify the domain,intent and entity recognition ***,training such intelligent system necessitates a large number of user-annotated datasets for each *** study proposes a single joint predictive deep neural network framework based on long short-term memory using only a small user-annotated dataset to address these *** investigates value added by incorporating unlabeled data from user chatting logs into multi-domain spoken language understanding *** experimental analysis of the proposed joint frameworks,along with the semi-supervised multi-domain model,using open-source annotated and unannotated utterances shows robust improvement in the predictive performance of the proposed multi-domain intelligent chatbot over a base joint model and joint model based on adversarial learning.

关键词： Chatbot dialog system joint learning LSTM natural language understanding semi-supervised learning

Data Augmentation Using Contour Image for Convolutional Neural Network

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Computers, Materials & Continua 2023年第6期75卷 4669-4680页

作者： Seung-Yeon Hwang jeong-Joon Kim Department of Computer Engineering Anyang UniversityAnyang-si14058Korea Department of ICT Convergence Engineering Anyang UniversityAnyang-si14058Korea

With the development of artificial intelligence-related technologies such as deep learning,various organizations,including the government,are making various efforts to generate and manage big data for use in artificial ***,it is difficult to acquire big data due to various social problems and restrictions such as personal information *** are many problems in introducing technology in fields that do not have enough training data necessary to apply deep learning ***,this study proposes a mixed contour data augmentation technique,which is a data augmentation technique using contour images,to solve a problem caused by a lack of ***,a famous convolutional neural network(CNN)architecture,and CIFAR-10,a benchmark data set,are used for experimental performance evaluation to prove the superiority of the proposed *** to prove that high performance improvement can be achieved even with a small training dataset,the ratio of the training dataset was divided into 70%,50%,and 30%for comparative *** a result of applying the mixed contour data augmentation technique,it was possible to achieve a classification accuracy improvement of up to 4.64%and high accuracy even with a small amount of data *** addition,it is expected that the mixed contour data augmentation technique can be applied in various fields by proving the excellence of the proposed data augmentation technique using benchmark datasets.

关键词： Data augmentation image classification deep learning convolutional neural network mixed contour image benchmark dataset

Multi-View Multi-Modal Head-Gaze Estimation for Advanced Indoor User Interaction

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Computers, Materials & Continua 2022年第3期70卷 5107-5132页

作者： Jung-Hwa Kim Jin-Woo jeong Department of Computer Engineering Kumoh National Institute of TechnologyGumi39177Korea Department of Data Science Seoul National University of Science and TechnologySeoul01811Kore

Gaze estimation is one of the most promising technologies for supporting indoor monitoring and interaction ***,previous gaze estimation techniques generally work only in a controlled laboratory environment because they require a number of high-resolution eye *** makes them unsuitable for welfare and healthcare facilities with the following challenging characteristics:1)users’continuous movements,2)various lighting conditions,and 3)a limited amount of available *** address these issues,we introduce a multi-view multi-modal head-gaze estimation system that translates the user’s head orientation into the gaze *** proposed system captures the user using multiple cameras with depth and infrared modalities to train more robust gaze estimators under the aforementioned *** this end,we implemented a deep learning pipeline that can handle different types and combinations of *** proposed system was evaluated using the data collected from 10 volunteer participants to analyze how the use of single/multiple cameras and modalities affect the performance of head-gaze *** various experiments,we found that 1)an infrared-modality provides more useful features than a depth-modality,2)multi-view multi-modal approaches provide better accuracy than singleview single-modal approaches,and 3)the proposed estimators achieve a high inference efficiency that can be used in real-time applications.

关键词： Human-computer interaction deep learning head-gaze estimation indoor monitoring

A Universal Activation Function for Deep Learning

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Computers, Materials & Continua 2023年第5期75卷 3553-3569页

作者： Seung-Yeon Hwang jeong-Joon Kim Department of Computer Engineering Anyang UniversityAnyang-si14028Korea Department of ICT Convergence Engineering Anyang UniversityAnyang-si14028Korea

Recently,deep learning has achieved remarkable results in fields that require human cognitive ability,learning ability,and reasoning *** functions are very important because they provide the ability of artificial neural networks to learn complex patterns through *** activation functions are being studied to solve problems such as vanishing gradients and dying nodes that may occur in the deep learning ***,it takes a lot of time and effort for researchers to use the existing activation function in their ***,in this paper,we propose a universal activation function(UA)so that researchers can easily create and apply various activation functions and improve the performance of neural *** can generate new types of activation functions as well as functions like traditional activation functions by properly adjusting three *** famous Convolutional Neural Network(CNN)and benchmark datasetwere used to evaluate the experimental performance of the UA proposed in this *** compared the performance of the artificial neural network to which the traditional activation function is applied and the artificial neural network to which theUA is *** addition,we evaluated the performance of the new activation function generated by adjusting the hyperparameters of *** experimental performance evaluation results showed that the classification performance of CNNs improved by up to 5%through the UA,although most of them showed similar performance to the traditional activation function.

关键词： Deep learning activation function convolutional neural network benchmark datasets universal activation function

A novel cytokine consisting of the p40 and EBI3 subunits suppresses experimental autoimmune arthritis via reciprocal regulation of Th17 and Treg cells

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Cellular & Molecular Immunology 2022年第1期19卷 79-91页

作者： Seon-Yeong Lee Su-Jin Moon Young-Mee Moon hyeon-Beom Seo Jun-Geol Ryu A Ram Lee Chae Rim Lee Da-Som Kim Yang-Mi Her jeong Won Choi Seung-Ki Kwok Sung-Hwan Park Mi-La Cho The Rheumatism Research Center The Catholic University of KoreaSeoulSouth Korea Division of Rheumatology Department of Internal MedicineSeoul St.Mary’s HospitalCollege of MedicineThe Catholic University of KoreaSeoulSouth Korea Department of Biomedicine&Health Sciences College of MedicineThe Catholic University of KoreaSeoulSouth Korea Department of Medical Lifescience College of MedicineThe Catholic University of KoreaSeoulSouth Korea These author contributed equally:Seon-Yeong Lee Su-Jin MoonYoung-Mee Moon

OBJECTIVE:The interleukin(IL)-12 cytokine family is closely related to the development of T helper cells,which are responsible for autoimmune disease enhancement or ***-12 family members are generally heterodimers and share threeα-subunits(p35,p19,and p28)and twoβ-subunits(p40 and EBI3).However,aβ-sheet p40 homodimer has been shown to exist and antagonize IL-12 and IL-23 signaling ***,we assumed the existence of a p40-EBI3 heterodimer in nature and sought to investigate its role in immune ***:The presence of the p40-EBI3 heterodimer was confirmed by ELISA,immunoprecipitation,and western blotting.A p40-EBI3 vector and p40-EBI3-Fc protein were synthesized to confirm the immunological role of this protein in mice with collageninduced arthritis(CIA).The anti-inflammatory effects of p40-EBI3 were analyzed with regard to clinical,histological,and immune cell-regulating features in mice with ***:Clinical arthritis scores and the expression levels of proinflammatory cytokines(e.g.,IL-17,IL-1β,IL-6,and TNF-α)were significantly attenuated in p40-EBI3-overexpressing and p40-EBI3-Fc-treated mice with CIA compared to vehicle-treated mice with *** joint damage and vessel formation-related gene expression were also reduced by p40-EBI3 heterodimer *** vitro,the p40-EBI3-Fc protein significantly suppressed the differentiation of Th17 cells and reciprocally induced CD4+CD25+Foxp3+(regulatory T)cells.p40-EBI3 also inhibited osteoclast formation in a concentrationdependent ***:In this study,p40-EBI3 ameliorated proinflammatory conditions both in vivo and in *** propose that p40-EBI3 is a novel anti-inflammatory cytokine involved in suppressing the immune response through the expansion of Treg cells and suppression of Th17 cells and osteoclastogenesis.

关键词： cytokine p40-EBI3 Rheumatoid arthritis Autoimmune disease regulatory T cell

Prognostic role of plasma level of angiopoietin-1,angiopoietin-2,and vascular endothelial growth factor in hepatocellular carcinoma

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

World Journal of Gastroenterology 2021年第27期27卷 4453-4467页

作者： Gwang hyeon Choi Eun Sun Jang Jin-Wook Kim Sook-Hyang jeong Department of Internal Medicine Seoul National University Bundang HospitalSeoul National University College of MedicineSeongnam 13620South Korea

BACKGROUND Most hepatocellular carcinomas(HCCs)are hypervascular,with characteristic features of hepatic arterial supply to the *** factors involved in tumor angiogenesis include angiopoietin-1(Ang-1),angiopoietin-2(Ang-2),and vascular endothelial growth factor(VEGF).AIM To investigate the profiles of plasma levels of angiogenesis markers in patients with HCC and evaluate their roles in predicting overall survival(OS)and progression-free survival(PFS).METHODS Plasma samples from 240 prospectively enrolled HCC patients in the very early to advanced stages were used to measure the levels of Ang-1,Ang-2,and *** associations with clinical characteristics,OS,and PFS were *** The median plasma levels of Ang-1,Ang-2,and VEGF were 3216 pg/mL,1684 pg/mL,and 26.5 pg/mL,*** plasma level of Ang-2 showed a significant increase from early stage[Barcelona clinic liver cancer(BCLC)A]to intermediate(BCLC B)and advanced stage HCC(BCLC C/D),whereas Ang-1,VEGF,and alpha-fetoprotein(AFP)levels in the plasma did not show any such *** analysis,propensity score-matched analysis,and timedependent receiver operating curve analysis revealed that Ang-2 levels had the highest predictive power for OS and *** Ang-1 nor VEGF was significantly associated with OS or *** neutrophil-to-lymphocyte ratio was an independent factor for OS and *** The plasma levels of Ang-2 correlated with liver function,tumor stage,and tumor invasiveness,showing better performance in predicting OS and PFS than AFP,Ang-1,or VEGF.

关键词： Hepatocellular carcinoma Angiogenesis Biomarker Prognosis Survival

Usage patterns of emergency medical services in Korea: analysis of patient flow

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Chinese Medical Journal 2019年第3期132卷 259-268页

作者： Mira Oh Jae hyeon Lee Byoung Uk Jeon Tae Oh jeong Tag Heo Sungmin Lee National Emergency Medical Center of National Medical Center Joong-guSeoulRepublic of Korea Department of Emergency medicine Chonbuk National University Medical SchoolDeokjin-guJeonjuRepublic of Korea Department of Emergency Medicine Chonnam National University Medical SchoolDong-guGwangjuRepublic of Korea

Background:This study used the National Emergency Department Information System (NEDIS) data to analyze the flow of emergency and critical emergency patients and to identify the patterns of emergency medical service usage in ***:The relevance index (RI) and commitment index (CI) were calculated from the 2016 NEDIS data. In this study, the number of clusters was determined using NbClust, and cluster analysis was used to analyze the usage patterns of emergency and critical emergency ***:The RI and CI were calculated using 8,389,766 cases of 214 districts. The results of the RI and CI suggested that there were 3 types of clusters among the emergency patients. In Cluster 1, 54 districts (25.2%) had low RI and high CI, and it was of outflow type. Cluster 2 was categorized as the influx-type in 58 districts (27.1%) irrespective of RI and low CI. Cluster 3 was categorized as the self-sufficient type found in 102 districts (47.7%), with high RI and high CI. The cluster analysis of the critical emergency patients was divided into 2 types. Cluster 1 was categorized as outflow type with high CI found in 129 districts (60.3%), while Cluster 2 was categorized as inflow type with low CI found in 85 districts (39.7%).Conclusions:This study elucidates the regional status of usage patterns of emergency and critical emergency patients in Korea. This study might serve as a basis for the establishment and selection of emergency medical service areas and vulnerable emergency medical service areas.

关键词： Clustering Commitment Index Emergency Department National Emergency Department Information System Relevance Index

Risk of ischaemic stroke in patients with transient global amnesia: a propensity-matched cohort study

维普期刊数据库

同方期刊数据库评论

在线全文

学校读者我要写书评

暂无评论

Stroke & Vascular Neurology 2022年第2期7卷 101-107页

作者： Sang Hun Lee Keon-Yeup Kim jeong-Woo Lee So-jeong Park Jin-Man Jung Department of Neurology Korea University Ansan HospitalKorea University College of MedicineAnsanRepublic of Korea Data Science Team Hanmi Pharm.Co.LtdSeoulRepublic of Korea Korea University Zebrafish Translational Medical Research Center AnsanRepublic of Korea

Background The exact pathophysiological mechanism of transient global amnesia(TGA)is *** is debatable whether TGA is a risk factor for ***,here we investigated the possibility of TGA as a risk factor for stroke in a real-world setting using large-scale nationwide health claims *** We used health claims data from the Korean National Health Insurance Service(NHIS).Patients diagnosed with TGA between 2007 and 2013 were *** initially extracted patients without TGA who were preferentially matched for age and sex with the patients with TGA at a ratio of 10:1 from the whole ***,we performed 1:2 propensity score matching analysis to balance the baseline characteristics between the two *** the propensity score-matched dataset,we performed multivariable Cox regression analysis to investigate the association between TGA and stroke type,including ischaemic,haemorrhagic and all stroke *** Patients with TGA(n=14673)were selected from the NHIS *** extracting from the whole database(n=140486)and propensity score matching their data at a 1:2 ratio,a total of 10448 and 20442 patients were finally assigned to the TGA and control groups,*** multivariable Cox regression analysis demonstrated that the TGA group had a higher risk of ischaemic stroke and all types of stroke(adjusted HR=1.194;95%CI:1.043 to 1.368;and HR=1.197;95%CI:1.056 to 1.357,respectively).Conclusions Analysis of the nationwide claims database showed that TGA could be an important risk factor for stroke,especially for ischaemic stroke.

关键词： patients matched transient