Pub Date : 2021-11-07DOI: 10.1109/CCIS53392.2021.9754670
Xiaoxiao Dong, Wei Wang, Haisheng Li, Qiang Cai
Recently, food segmentation has obtained growing attention in the field of computer vision for its great potential in human health. Most of existing methods utilize deep visual features extracting from Convolutional Neural Networks (CNNs) for food segmentation. However, these works ignore characteristics of food images and are thus difficult to achieve optimal segmentation performance. Compared with general image segmentation, food images usually do not exhibit unique spatial layout and common semantic patterns. In this paper, we address the food image segmentation task by capturing richer contextual and boundary information. The previous works capture image representation by multi-scale feature fusion, we propose a Windows Attention based Pyramid Network (WAPNet) to adaptively combine local features with global dependencies. Specifically, WAPNet combines Feature Pyramid Network (FPN) with Window Attention to weight multi-scale features, and then extract richer marginal information. In addition, we utilize a multimodality pre-training approach Recipe Learning Module (ReLeM) that explicitly provides segmentation model with rich semantic food knowledge. And by introducing Locality and Windows design, calculating self-attention according to Windows, We demonstrate promising performance on a new proposed food image benchmark for semantic segmentation.
{"title":"Windows Attention Based Pyramid Network for Food Segmentation","authors":"Xiaoxiao Dong, Wei Wang, Haisheng Li, Qiang Cai","doi":"10.1109/CCIS53392.2021.9754670","DOIUrl":"https://doi.org/10.1109/CCIS53392.2021.9754670","url":null,"abstract":"Recently, food segmentation has obtained growing attention in the field of computer vision for its great potential in human health. Most of existing methods utilize deep visual features extracting from Convolutional Neural Networks (CNNs) for food segmentation. However, these works ignore characteristics of food images and are thus difficult to achieve optimal segmentation performance. Compared with general image segmentation, food images usually do not exhibit unique spatial layout and common semantic patterns. In this paper, we address the food image segmentation task by capturing richer contextual and boundary information. The previous works capture image representation by multi-scale feature fusion, we propose a Windows Attention based Pyramid Network (WAPNet) to adaptively combine local features with global dependencies. Specifically, WAPNet combines Feature Pyramid Network (FPN) with Window Attention to weight multi-scale features, and then extract richer marginal information. In addition, we utilize a multimodality pre-training approach Recipe Learning Module (ReLeM) that explicitly provides segmentation model with rich semantic food knowledge. And by introducing Locality and Windows design, calculating self-attention according to Windows, We demonstrate promising performance on a new proposed food image benchmark for semantic segmentation.","PeriodicalId":191226,"journal":{"name":"2021 IEEE 7th International Conference on Cloud Computing and Intelligent Systems (CCIS)","volume":"19 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115184673","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-11-07DOI: 10.1109/CCIS53392.2021.9754665
Haoquan Fang, Lei Fan, Jenq-Neng Hwang
Autism spectrum disorder (ASD) is a comprehensive mental development disorder characterized by abnormal interpersonal communication and interaction patterns, narrow scope of interests, and limited content of activities. Due to the lack of biological diagnostic indicators, the current diagnosis of ASD mainly relies on experts’ comprehensive clinical analysis of children, which is usually subjective and highly dependent on doctors’ individual professional skills. In this study, we propose an auxiliary diagnostic method for early ASD, which is based on the eye movement data analysis of autistic children. The method involves biological motion visualization, eye tracking, machine learning, and other related techniques. More specifically, the visualized biological motion animation is divided into five stages according to different biological behaviors of human skeletal figures presented in the animation. At the same time, the screen is divided into six areas to represent different regions the children gaze at. Following these two temporal and spatial principles, features can be extracted from eye movement data. Based on those extracted features, machine learning methods, particularly KNN, Gaussian-NB, and Cubic-SVM, are trained to classify and diagnose autistic children, making future timely treatment possible.
{"title":"Auxiliary Diagnostic Method for Early Autism Spectrum Disorder Based on Eye Movement Data Analysis","authors":"Haoquan Fang, Lei Fan, Jenq-Neng Hwang","doi":"10.1109/CCIS53392.2021.9754665","DOIUrl":"https://doi.org/10.1109/CCIS53392.2021.9754665","url":null,"abstract":"Autism spectrum disorder (ASD) is a comprehensive mental development disorder characterized by abnormal interpersonal communication and interaction patterns, narrow scope of interests, and limited content of activities. Due to the lack of biological diagnostic indicators, the current diagnosis of ASD mainly relies on experts’ comprehensive clinical analysis of children, which is usually subjective and highly dependent on doctors’ individual professional skills. In this study, we propose an auxiliary diagnostic method for early ASD, which is based on the eye movement data analysis of autistic children. The method involves biological motion visualization, eye tracking, machine learning, and other related techniques. More specifically, the visualized biological motion animation is divided into five stages according to different biological behaviors of human skeletal figures presented in the animation. At the same time, the screen is divided into six areas to represent different regions the children gaze at. Following these two temporal and spatial principles, features can be extracted from eye movement data. Based on those extracted features, machine learning methods, particularly KNN, Gaussian-NB, and Cubic-SVM, are trained to classify and diagnose autistic children, making future timely treatment possible.","PeriodicalId":191226,"journal":{"name":"2021 IEEE 7th International Conference on Cloud Computing and Intelligent Systems (CCIS)","volume":"27 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122888242","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-11-07DOI: 10.1109/CCIS53392.2021.9754605
Zheng Cao, F. Meng
An adaptive observer-based Inverse optimal controller (AOC) is proposed for a class of second-order Euler-Lagrange systems with various uncertainties in the dynamic models. Specifically, the proposed AOC adopts one NN-based robust adaptive inverse optimal controller to approximate the nonlinear unknown system and generate optimal control inputs, while the other NN-based adaptive observer is established to estimate the unmeasured system state. The developed AOC is proved to achieve semi-global asymptotic optimal tracking (by inverse optimal controller) through Lyapunov stability analysis. Simulation analysis shows that the AOC has small tracking error even with the observed information in the presence of uncertain disturbances.
{"title":"Adaptive Observer-Based Inverse Optimal Control of a Class of Second-Order Euler-Lagrange Systems","authors":"Zheng Cao, F. Meng","doi":"10.1109/CCIS53392.2021.9754605","DOIUrl":"https://doi.org/10.1109/CCIS53392.2021.9754605","url":null,"abstract":"An adaptive observer-based Inverse optimal controller (AOC) is proposed for a class of second-order Euler-Lagrange systems with various uncertainties in the dynamic models. Specifically, the proposed AOC adopts one NN-based robust adaptive inverse optimal controller to approximate the nonlinear unknown system and generate optimal control inputs, while the other NN-based adaptive observer is established to estimate the unmeasured system state. The developed AOC is proved to achieve semi-global asymptotic optimal tracking (by inverse optimal controller) through Lyapunov stability analysis. Simulation analysis shows that the AOC has small tracking error even with the observed information in the presence of uncertain disturbances.","PeriodicalId":191226,"journal":{"name":"2021 IEEE 7th International Conference on Cloud Computing and Intelligent Systems (CCIS)","volume":"50 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128471272","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-11-07DOI: 10.1109/CCIS53392.2021.9754680
Fujian Feng, Han Huang, Yihui Liang
Image matting is an essential image processing technology. optimized-based image matting methods can significantly improve the alpha matte quality of high-resolution images. However, the local information of the foreground may be similar to the background, which causes the inversion problem of the alpha matte in the single-point optimized. In this paper, we propose an image matting mathematical model of the equal-in-space distance. The model transforms the high-resolution image matting problem into several small-scale combinatorial optimization problems according to the similarity among pixel features. Inspired by spanning tree, we propose a graph-order optimization strategy, which generates the optimization sequence of small-scale optimization problems according to the edge weight among graph nodes. In addition, we designed a graph-order optimization algorithm based on optimized information transfer to solve each node in the graph. Experimental results show that the proposed model solves the alpha matte inversion problem of single-point optimization matting. Besides, the proposed algorithm outperforms the state-of-the-art optimization algorithms for the high-resolution image matting problem.
{"title":"Graph-Order Optimization Algorithm Based on Equal-in-Space Distance Model for High-Resolution Image Matting","authors":"Fujian Feng, Han Huang, Yihui Liang","doi":"10.1109/CCIS53392.2021.9754680","DOIUrl":"https://doi.org/10.1109/CCIS53392.2021.9754680","url":null,"abstract":"Image matting is an essential image processing technology. optimized-based image matting methods can significantly improve the alpha matte quality of high-resolution images. However, the local information of the foreground may be similar to the background, which causes the inversion problem of the alpha matte in the single-point optimized. In this paper, we propose an image matting mathematical model of the equal-in-space distance. The model transforms the high-resolution image matting problem into several small-scale combinatorial optimization problems according to the similarity among pixel features. Inspired by spanning tree, we propose a graph-order optimization strategy, which generates the optimization sequence of small-scale optimization problems according to the edge weight among graph nodes. In addition, we designed a graph-order optimization algorithm based on optimized information transfer to solve each node in the graph. Experimental results show that the proposed model solves the alpha matte inversion problem of single-point optimization matting. Besides, the proposed algorithm outperforms the state-of-the-art optimization algorithms for the high-resolution image matting problem.","PeriodicalId":191226,"journal":{"name":"2021 IEEE 7th International Conference on Cloud Computing and Intelligent Systems (CCIS)","volume":"157 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128345036","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-11-07DOI: 10.1109/CCIS53392.2021.9754604
Minyan Zheng, Jianping Luo, Wenming Cao
In this paper, we proposed a Spatial-Temporal Transformer (STTF) algorithm for video super resolution (SR), to solve the problem of blurs or artifacts after super resolve low-resolution (LR) video with traditional super resolution algorithm. Firstly, the algorithm uses residual blocks to extract initial features from video sequences. Secondly, the three-dimensional video features are decomposed into image patches and then are sent to the Spatial-Temporal Transformer network for self-attention among patches where patches can be aligned and fused. Finally, sub-pixel convolution layer and residual layers are applied to up-sampling and reconstruct the high-resolution (HR) video sequences. In order to improve video visual effects, minimum mean square error (MSE) loss function is applied to train the neural network. The experimental results show that the STTF network has a higher peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM) compared to traditional super-resolution algorithm.
{"title":"Video Super-Resolution Based on Spatial-Temporal Transformer","authors":"Minyan Zheng, Jianping Luo, Wenming Cao","doi":"10.1109/CCIS53392.2021.9754604","DOIUrl":"https://doi.org/10.1109/CCIS53392.2021.9754604","url":null,"abstract":"In this paper, we proposed a Spatial-Temporal Transformer (STTF) algorithm for video super resolution (SR), to solve the problem of blurs or artifacts after super resolve low-resolution (LR) video with traditional super resolution algorithm. Firstly, the algorithm uses residual blocks to extract initial features from video sequences. Secondly, the three-dimensional video features are decomposed into image patches and then are sent to the Spatial-Temporal Transformer network for self-attention among patches where patches can be aligned and fused. Finally, sub-pixel convolution layer and residual layers are applied to up-sampling and reconstruct the high-resolution (HR) video sequences. In order to improve video visual effects, minimum mean square error (MSE) loss function is applied to train the neural network. The experimental results show that the STTF network has a higher peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM) compared to traditional super-resolution algorithm.","PeriodicalId":191226,"journal":{"name":"2021 IEEE 7th International Conference on Cloud Computing and Intelligent Systems (CCIS)","volume":"34 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124171273","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-11-07DOI: 10.1109/CCIS53392.2021.9754640
Changlong He, Zengyang Shao, Lijia Ma, Jianqiang Li, Tingyi Hu
The signed network has received widespread attention because it can well reflect the cooperation and conflict relationship. Structural balance is an important global feature in signed networks, which can well reflect the structural characteristics of the network. Existing structural balance calculation algorithms define the global and local balance computation problems as an optimization problem, and then optimize their respective objective functions through optimization algorithms, but these algorithms ignore the correlation between the two problems. In this paper, we combine the multifactorial evolutionary algorithm and the discrete particle swarm optimization algorithm, and further propose the multifactorial discrete particle swarm optimization algorithm (MFDPSO). This algorithm designs the knowledge transfer function and optimization algorithm based on the correlation of the strong and weak structure balance and optimizes the two problems at the same time. The experimental results on 8 real networks demonstrate the effectiveness of the MFDPSO.
{"title":"Structural Balance Computation in Signed Networks by Using Multifactorial Discrete Particle Swarm Optimization","authors":"Changlong He, Zengyang Shao, Lijia Ma, Jianqiang Li, Tingyi Hu","doi":"10.1109/CCIS53392.2021.9754640","DOIUrl":"https://doi.org/10.1109/CCIS53392.2021.9754640","url":null,"abstract":"The signed network has received widespread attention because it can well reflect the cooperation and conflict relationship. Structural balance is an important global feature in signed networks, which can well reflect the structural characteristics of the network. Existing structural balance calculation algorithms define the global and local balance computation problems as an optimization problem, and then optimize their respective objective functions through optimization algorithms, but these algorithms ignore the correlation between the two problems. In this paper, we combine the multifactorial evolutionary algorithm and the discrete particle swarm optimization algorithm, and further propose the multifactorial discrete particle swarm optimization algorithm (MFDPSO). This algorithm designs the knowledge transfer function and optimization algorithm based on the correlation of the strong and weak structure balance and optimizes the two problems at the same time. The experimental results on 8 real networks demonstrate the effectiveness of the MFDPSO.","PeriodicalId":191226,"journal":{"name":"2021 IEEE 7th International Conference on Cloud Computing and Intelligent Systems (CCIS)","volume":"56 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124354577","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-11-07DOI: 10.1109/CCIS53392.2021.9754621
Xiaoting Yu, Cao Liang, Hongyuan Wang, Suolan Liu, Yan Hui
At present, supervised video-based person re-identification has achieved excellent performance. However, the initial video data obtained from real scenes are often unlabeled. Labelling such data is very time-consuming. If unsupervised learning can be effectively applied to these data, so much cost will be saved. In this paper, based on the joint global and local metric, an unsupervised video-based person re-identification method is proposed, which takes both the global information of a video sequence and the local information between the video frames into account to better distinguish different appearances of the same pedestrian. The global similarity and local similarity are calculated using global and local features, respectively. Meanwhile, a diversity constraint is used as an aid for cluster merging and evaluation. In the training process, the network is optimized by combining cluster mutual exclusion loss and center loss, which reduces the within-class differences and enlarges the between-class differences. Experiments on two benchmark datasets, MARS and DukMTMC-VideoReID, the results show that this method has higher accuracy and stabilityshow that the proposed method can achieve higher accuracy and is more stable than most state-of-the-art unsupervised methods.
{"title":"Unsupervised Video-based Person Re-identification Based on The Joint Global-local Metrics","authors":"Xiaoting Yu, Cao Liang, Hongyuan Wang, Suolan Liu, Yan Hui","doi":"10.1109/CCIS53392.2021.9754621","DOIUrl":"https://doi.org/10.1109/CCIS53392.2021.9754621","url":null,"abstract":"At present, supervised video-based person re-identification has achieved excellent performance. However, the initial video data obtained from real scenes are often unlabeled. Labelling such data is very time-consuming. If unsupervised learning can be effectively applied to these data, so much cost will be saved. In this paper, based on the joint global and local metric, an unsupervised video-based person re-identification method is proposed, which takes both the global information of a video sequence and the local information between the video frames into account to better distinguish different appearances of the same pedestrian. The global similarity and local similarity are calculated using global and local features, respectively. Meanwhile, a diversity constraint is used as an aid for cluster merging and evaluation. In the training process, the network is optimized by combining cluster mutual exclusion loss and center loss, which reduces the within-class differences and enlarges the between-class differences. Experiments on two benchmark datasets, MARS and DukMTMC-VideoReID, the results show that this method has higher accuracy and stabilityshow that the proposed method can achieve higher accuracy and is more stable than most state-of-the-art unsupervised methods.","PeriodicalId":191226,"journal":{"name":"2021 IEEE 7th International Conference on Cloud Computing and Intelligent Systems (CCIS)","volume":"29 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131994922","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-11-07DOI: 10.1109/CCIS53392.2021.9754634
Mingyu Li, Yujing Liu, Zhengsen Ji, D. Niu, Huanfen Zhang
At present, the power load of large power users such as electric power manufacturing enterprises is greatly affected by abnormal factors, among which the weather factor is one of the important influencing factors. How to accurately forecast the load level by considering weather factors is of great significance. This paper uses cluster analysis to screen out similar days that are severely affected by weather from the load data throughout the year. And a deep learning forecasting model that considers weather factors is built to realize the daily load forecast of electric power manufacturing enterprises. The realization of this research is helpful to provide accurate load forecasting methods for electric power manufacturing enterprises. The production plans according to weather conditions can be adjusted and the risks can be avoided, which can improve production efficiency.
{"title":"Daily Load Forecasting of Electric Power Manufacturing Industry Considering Disaster Weather Recognition Under the Deep Learning","authors":"Mingyu Li, Yujing Liu, Zhengsen Ji, D. Niu, Huanfen Zhang","doi":"10.1109/CCIS53392.2021.9754634","DOIUrl":"https://doi.org/10.1109/CCIS53392.2021.9754634","url":null,"abstract":"At present, the power load of large power users such as electric power manufacturing enterprises is greatly affected by abnormal factors, among which the weather factor is one of the important influencing factors. How to accurately forecast the load level by considering weather factors is of great significance. This paper uses cluster analysis to screen out similar days that are severely affected by weather from the load data throughout the year. And a deep learning forecasting model that considers weather factors is built to realize the daily load forecast of electric power manufacturing enterprises. The realization of this research is helpful to provide accurate load forecasting methods for electric power manufacturing enterprises. The production plans according to weather conditions can be adjusted and the risks can be avoided, which can improve production efficiency.","PeriodicalId":191226,"journal":{"name":"2021 IEEE 7th International Conference on Cloud Computing and Intelligent Systems (CCIS)","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131234577","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-11-07DOI: 10.1109/CCIS53392.2021.9754623
Xiaomeng Chen, Yingxia Shao, Zhe Xue, Ziqiang Yu
An effective and accurate method of detecting COVID-19 infection is to analyze medical diagnostic images (e.g. CT scans). However, patients’ information is privacy, and it is illegal to share diagnostic images among medical institutions. In this case, a critical issue faced by the model that detects the CT images is lacking enough training images dataset, then the features of COVID-19 cannot be accurately obtained. The data privacy attracts extensive attentions recently and is particularly important for the fast-developing medical institution database and. Considering this point, this paper presents a blockchain federated learning model, which overcomes the burden of centralized collection of large amounts of sensitive data. The model uses a trained model to recognize CT scans, and shares data between hospitals with privacy protection mechanism. This model is able to learn from shared resources or data between different hospital repositories to discover patients with new coronary pneumonia by detecting the computed tomography (CT) images. Finally, we conduct extensive experiments to verify the performance of the model.
{"title":"Multi-Modal COVID-19 Discovery With Collaborative Federated Learning","authors":"Xiaomeng Chen, Yingxia Shao, Zhe Xue, Ziqiang Yu","doi":"10.1109/CCIS53392.2021.9754623","DOIUrl":"https://doi.org/10.1109/CCIS53392.2021.9754623","url":null,"abstract":"An effective and accurate method of detecting COVID-19 infection is to analyze medical diagnostic images (e.g. CT scans). However, patients’ information is privacy, and it is illegal to share diagnostic images among medical institutions. In this case, a critical issue faced by the model that detects the CT images is lacking enough training images dataset, then the features of COVID-19 cannot be accurately obtained. The data privacy attracts extensive attentions recently and is particularly important for the fast-developing medical institution database and. Considering this point, this paper presents a blockchain federated learning model, which overcomes the burden of centralized collection of large amounts of sensitive data. The model uses a trained model to recognize CT scans, and shares data between hospitals with privacy protection mechanism. This model is able to learn from shared resources or data between different hospital repositories to discover patients with new coronary pneumonia by detecting the computed tomography (CT) images. Finally, we conduct extensive experiments to verify the performance of the model.","PeriodicalId":191226,"journal":{"name":"2021 IEEE 7th International Conference on Cloud Computing and Intelligent Systems (CCIS)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133697048","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-11-07DOI: 10.1109/CCIS53392.2021.9754663
Suyu Ouyang, Junping Du, Benzhi Wang, Bowen Yu, Yuhui Wang, M. Liang
As social media becomes more and more popular, fake news spreads rapidly which is more likely to cause serious consequences, especially during the COVID-19 pandemic. On the premise of meeting data privacy and security requirements, federated learning uses multi-party heterogeneous data to further promote machine learning. This paper proposes a federal learning based COVID-19 fake news detection model with deep self-attention network (FL_FNDM). We construct a deep self-attention network for fake news detection, which combines self-attention-based pretrained model BERT and deep convolutional neural network to detect fake news. Moreover, the fake news detection model is learned under the framework of horizontal federated learning, aiming at protecting users’ data security and privacy. The experimental results demonstrate that the proposed model can improve the performance of fake news detection on the COVID-19 dataset, which can achieve almost the same effect of sharing data without leaking user data.
{"title":"Federal Learning Based COVID-19 Fake News Detection With Deep Self-Attention Network","authors":"Suyu Ouyang, Junping Du, Benzhi Wang, Bowen Yu, Yuhui Wang, M. Liang","doi":"10.1109/CCIS53392.2021.9754663","DOIUrl":"https://doi.org/10.1109/CCIS53392.2021.9754663","url":null,"abstract":"As social media becomes more and more popular, fake news spreads rapidly which is more likely to cause serious consequences, especially during the COVID-19 pandemic. On the premise of meeting data privacy and security requirements, federated learning uses multi-party heterogeneous data to further promote machine learning. This paper proposes a federal learning based COVID-19 fake news detection model with deep self-attention network (FL_FNDM). We construct a deep self-attention network for fake news detection, which combines self-attention-based pretrained model BERT and deep convolutional neural network to detect fake news. Moreover, the fake news detection model is learned under the framework of horizontal federated learning, aiming at protecting users’ data security and privacy. The experimental results demonstrate that the proposed model can improve the performance of fake news detection on the COVID-19 dataset, which can achieve almost the same effect of sharing data without leaking user data.","PeriodicalId":191226,"journal":{"name":"2021 IEEE 7th International Conference on Cloud Computing and Intelligent Systems (CCIS)","volume":"40 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128846732","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：