Pub Date : 2020-06-01DOI: 10.1109/ISCV49265.2020.9204099
M. Qjidaa, Y. Mechbal, A. Ben-fares, H. Amakdouf, M. Maaroufi, B. Alami, H. Qjidaa
To combat the spread of COVID 19, the World Health Organization suggests a large-scale implementation of COVID 19 tests. Unfortunately, these tests are expensive and cannot be provided and available for people in rural and remote areas. To remedy this problem, we will develop an intelligent clinical decision support system (SADC) for the early diagnosis of COVID 19 from chest x-rays which are more accessible for people in rural areas. Thus, we collected a total of 566 radiological images classified into 3 classes: a class of COVID19 type, a Class of Pneumonia type and a class of Normal type. In the experimental analysis, 70% of the data set was used as training set and 30% was used as the test set. After preprocessing process, we use some augmentation using a rotation, a horizontal flip, a channel shift and rescale. Our finale classifier achieved the best performance with test accuracy of 99%, f1score 98%, precision of 98.60% and sensitivity 98.30%.
{"title":"Early detection of COVID19 by deep learning transfer Model for populations in isolated rural areas","authors":"M. Qjidaa, Y. Mechbal, A. Ben-fares, H. Amakdouf, M. Maaroufi, B. Alami, H. Qjidaa","doi":"10.1109/ISCV49265.2020.9204099","DOIUrl":"https://doi.org/10.1109/ISCV49265.2020.9204099","url":null,"abstract":"To combat the spread of COVID 19, the World Health Organization suggests a large-scale implementation of COVID 19 tests. Unfortunately, these tests are expensive and cannot be provided and available for people in rural and remote areas. To remedy this problem, we will develop an intelligent clinical decision support system (SADC) for the early diagnosis of COVID 19 from chest x-rays which are more accessible for people in rural areas. Thus, we collected a total of 566 radiological images classified into 3 classes: a class of COVID19 type, a Class of Pneumonia type and a class of Normal type. In the experimental analysis, 70% of the data set was used as training set and 30% was used as the test set. After preprocessing process, we use some augmentation using a rotation, a horizontal flip, a channel shift and rescale. Our finale classifier achieved the best performance with test accuracy of 99%, f1score 98%, precision of 98.60% and sensitivity 98.30%.","PeriodicalId":313743,"journal":{"name":"2020 International Conference on Intelligent Systems and Computer Vision (ISCV)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124858077","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 : 2020-06-01DOI: 10.1109/ISCV49265.2020.9204217
Said Radouche, C. Leghris
In the next generation heterogeneous wireless and mobile networks, the mobile terminal equipped with a multi-interface may be able to choose an optimal access network anywhere and at any time. The seamless handover between different technologies (Cellular, WiMAX, Wi-Fi, and Satellite) is referred to as vertical handover (VHO). However, the main challenge is to provide seamless connectivity to the mobile terminal in this heterogeneous environment. Therefore, VHO needs an effective network selection process based on multiple network parameters. This article presents a new network selection algorithm based on cosine similarity to rank alternative networks and combination method that integrates both subjective weights, calculated using the user’s experience, and objective weights, determined by the Entropy method. The performance indicators used in this study are ranking abnormality and the number of handoffs. Obtained results indicate that the developed method performed better than the conventional MADM methods widely used in the context of vertical handover namely TOPSIS, VIKOR, and GRA.
{"title":"Network selection based on Cosine Similarity and Combination of Subjective and Objective Weighting","authors":"Said Radouche, C. Leghris","doi":"10.1109/ISCV49265.2020.9204217","DOIUrl":"https://doi.org/10.1109/ISCV49265.2020.9204217","url":null,"abstract":"In the next generation heterogeneous wireless and mobile networks, the mobile terminal equipped with a multi-interface may be able to choose an optimal access network anywhere and at any time. The seamless handover between different technologies (Cellular, WiMAX, Wi-Fi, and Satellite) is referred to as vertical handover (VHO). However, the main challenge is to provide seamless connectivity to the mobile terminal in this heterogeneous environment. Therefore, VHO needs an effective network selection process based on multiple network parameters. This article presents a new network selection algorithm based on cosine similarity to rank alternative networks and combination method that integrates both subjective weights, calculated using the user’s experience, and objective weights, determined by the Entropy method. The performance indicators used in this study are ranking abnormality and the number of handoffs. Obtained results indicate that the developed method performed better than the conventional MADM methods widely used in the context of vertical handover namely TOPSIS, VIKOR, and GRA.","PeriodicalId":313743,"journal":{"name":"2020 International Conference on Intelligent Systems and Computer Vision (ISCV)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116559089","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 : 2020-06-01DOI: 10.1109/ISCV49265.2020.9204138
Imane El Boujnouni, Abdelhak Tali, K. Bentaleb
Myocardial infarction (MI) is one of the leading causes of mortality throughout the world. Early diagnosis of MI is crucial for effective treatment to avoid patient morality. In this regard, the most commonly used technique for the problem of MI detection is the Convolutional Neural Network (CNN), which has shown good performance, but it still has some limitations. CNN requires a large amount of data, which is a challenge in the medical field. Therefore, the proposed approach uses a novel architecture consisting of wavelet transform and Capsule network, which is the most advanced algorithm to overcome CNN’s drawback. Experimental results achieve an accuracy of 91.2%, Sensitivity of 83% and Specificity of 89.5% which demonstrates that CapsNet acquires promising results while using fewer data.
{"title":"Capsule Network Based on Scalograms of Electrocardiogram for Myocardial Infarction Classification","authors":"Imane El Boujnouni, Abdelhak Tali, K. Bentaleb","doi":"10.1109/ISCV49265.2020.9204138","DOIUrl":"https://doi.org/10.1109/ISCV49265.2020.9204138","url":null,"abstract":"Myocardial infarction (MI) is one of the leading causes of mortality throughout the world. Early diagnosis of MI is crucial for effective treatment to avoid patient morality. In this regard, the most commonly used technique for the problem of MI detection is the Convolutional Neural Network (CNN), which has shown good performance, but it still has some limitations. CNN requires a large amount of data, which is a challenge in the medical field. Therefore, the proposed approach uses a novel architecture consisting of wavelet transform and Capsule network, which is the most advanced algorithm to overcome CNN’s drawback. Experimental results achieve an accuracy of 91.2%, Sensitivity of 83% and Specificity of 89.5% which demonstrates that CapsNet acquires promising results while using fewer data.","PeriodicalId":313743,"journal":{"name":"2020 International Conference on Intelligent Systems and Computer Vision (ISCV)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128533772","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 : 2020-06-01DOI: 10.1109/ISCV49265.2020.9204033
Amal Hammani, Hamza Hamout, A. Elyousfi
3D-HEVC is developed by ISO/IEC MPEG and ITU-T Video Coding Experts Group (VCEG) as the highestprofile extension of HEVC for 3D video coding and Multi-View texture Videos plus Depth maps (MVD) compress. Concerning the current 3D-HEVC design, the uniform HEVC Intra prediction and Depth Modeling Modes (DMMs) are used in a sophisticated way to ameliorate the performance of the depth map video coding. Although, the enhancement of the coding efficiency is achieved at the cost of computational complexity load, which prevents 3D-HEVC from being used in real-world applications. Thus, we suggest a fast depth map Intra prediction mode decision based on the Self-organizing Map to resolve the aforementioned computational complexity increases. The experimental results demonstrate that the suggested algorithm can increase the encoding time savings up to 39.8%, with insignificant rate-distortion loss.
{"title":"Fast Depth Map Intra Mode Prediction Based on Self-organizing Map","authors":"Amal Hammani, Hamza Hamout, A. Elyousfi","doi":"10.1109/ISCV49265.2020.9204033","DOIUrl":"https://doi.org/10.1109/ISCV49265.2020.9204033","url":null,"abstract":"3D-HEVC is developed by ISO/IEC MPEG and ITU-T Video Coding Experts Group (VCEG) as the highestprofile extension of HEVC for 3D video coding and Multi-View texture Videos plus Depth maps (MVD) compress. Concerning the current 3D-HEVC design, the uniform HEVC Intra prediction and Depth Modeling Modes (DMMs) are used in a sophisticated way to ameliorate the performance of the depth map video coding. Although, the enhancement of the coding efficiency is achieved at the cost of computational complexity load, which prevents 3D-HEVC from being used in real-world applications. Thus, we suggest a fast depth map Intra prediction mode decision based on the Self-organizing Map to resolve the aforementioned computational complexity increases. The experimental results demonstrate that the suggested algorithm can increase the encoding time savings up to 39.8%, with insignificant rate-distortion loss.","PeriodicalId":313743,"journal":{"name":"2020 International Conference on Intelligent Systems and Computer Vision (ISCV)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129033985","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 : 2020-06-01DOI: 10.1109/ISCV49265.2020.9204168
Chama El Saili, Soukaina Fatimi, L. Alaoui
Text clustering is a crucial application of data mining. It can be used to structure hypertext documents or large sets of text. Many research works have dived into document clustering as a technique for improving search, information retrieval, document browsing, automatic topic identification, as well as the primitive task of clustering. Major challenges are entangling researchers, especially when working with large scale datasets, such as very high dimensionality and cluster labeling. To tackle these challenges, a number of techniques using frequent itemsets mining methods in text clustering have been proposed. In this paper, we review such techniques while highlighting their strengths and limitations. With the analysis of associated methodologies, we also propose a general framework for the task of text clustering using frequent itemsets mining algorithms.
{"title":"Frequent Itemsets Methods for Text Clustering","authors":"Chama El Saili, Soukaina Fatimi, L. Alaoui","doi":"10.1109/ISCV49265.2020.9204168","DOIUrl":"https://doi.org/10.1109/ISCV49265.2020.9204168","url":null,"abstract":"Text clustering is a crucial application of data mining. It can be used to structure hypertext documents or large sets of text. Many research works have dived into document clustering as a technique for improving search, information retrieval, document browsing, automatic topic identification, as well as the primitive task of clustering. Major challenges are entangling researchers, especially when working with large scale datasets, such as very high dimensionality and cluster labeling. To tackle these challenges, a number of techniques using frequent itemsets mining methods in text clustering have been proposed. In this paper, we review such techniques while highlighting their strengths and limitations. With the analysis of associated methodologies, we also propose a general framework for the task of text clustering using frequent itemsets mining algorithms.","PeriodicalId":313743,"journal":{"name":"2020 International Conference on Intelligent Systems and Computer Vision (ISCV)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129029657","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 : 2020-06-01DOI: 10.1109/ISCV49265.2020.9204249
Saloua El Motaki, Ali Yahyaouy, H. Gualous, J. Sabor
In this paper, a new weighted fuzzy c-means clustering algorithm is proposed. The presented approach consists of emulating the collective behaviour of starling birds to form homogeneous and well-separated clusters. In a flock of starlings, each individual maintains a connection with its neighborhood to determine its position in space. This connection allows the individual to approach the flock-mates that have the same direction as its own, and simultaneously, to move away from other individuals that have a different direction. Based on this metaphor, in this work, we propose the use of the three elementary movements of the starling bird, separation, alignment, and cohesion, to update the weight parameter associated with each individual in the dataset. The accuracy of the proposed algorithm has been assessed by two clustering validation indices. Compared to some existing algorithms, our algorithm provides better results. An example of artificial data is used to contrast some cases of this approach.
{"title":"A new weighted fuzzy c-means based on the collective behaviour of starling birds","authors":"Saloua El Motaki, Ali Yahyaouy, H. Gualous, J. Sabor","doi":"10.1109/ISCV49265.2020.9204249","DOIUrl":"https://doi.org/10.1109/ISCV49265.2020.9204249","url":null,"abstract":"In this paper, a new weighted fuzzy c-means clustering algorithm is proposed. The presented approach consists of emulating the collective behaviour of starling birds to form homogeneous and well-separated clusters. In a flock of starlings, each individual maintains a connection with its neighborhood to determine its position in space. This connection allows the individual to approach the flock-mates that have the same direction as its own, and simultaneously, to move away from other individuals that have a different direction. Based on this metaphor, in this work, we propose the use of the three elementary movements of the starling bird, separation, alignment, and cohesion, to update the weight parameter associated with each individual in the dataset. The accuracy of the proposed algorithm has been assessed by two clustering validation indices. Compared to some existing algorithms, our algorithm provides better results. An example of artificial data is used to contrast some cases of this approach.","PeriodicalId":313743,"journal":{"name":"2020 International Conference on Intelligent Systems and Computer Vision (ISCV)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116300180","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 : 2020-06-01DOI: 10.1109/ISCV49265.2020.9204293
Amina Bengag, A. Bengag, Mohamed el Boukhari
The main goal of VANET is to improve the traffic safety, by minimizing the number of road accidents. Hence, a reliable and rapid communication is crucial for vehicles. However, the high dynamic topology of VANETs, the frequently link breakage problem and the high speed of vehicles have brought great challenges to the routing design of VANETs. It is difficult for existing routing protocols for WSN and MANETs to adapt the high dynamics of vehicular networks. Therefore, various number of routing protocols have been proposed to deal with the specific characteristics of VANETs. GPSR for Greedy Perimeter Stateless Routing is the most popular protocol, but it still suffers from frequent link breakages issue due to the high-mobility of vehicles, which cause a low PDR and throughput. In this investigation, we introduce two new strategies to enhance the classical GPSR protocol, by reducing the problem of link breakages and getting a stable route that improve the PDR and throughput in addition to reduce the routing overhead. The two proposed routing protocols E-GPSR and DVA-GPSR guide the selection of the next hop node based not only on the position but also on other important metrics of the participating nodes. Simulation results show that DVA-GPSR and E-GPSR protocols perform remarkably GPSR in terms of PDR, throughput, and routing overhead by varying the number of vehicles in a real urban scenario.
{"title":"Enhancing GPSR routing protocol based on Velocity and Density for real-time urban scenario","authors":"Amina Bengag, A. Bengag, Mohamed el Boukhari","doi":"10.1109/ISCV49265.2020.9204293","DOIUrl":"https://doi.org/10.1109/ISCV49265.2020.9204293","url":null,"abstract":"The main goal of VANET is to improve the traffic safety, by minimizing the number of road accidents. Hence, a reliable and rapid communication is crucial for vehicles. However, the high dynamic topology of VANETs, the frequently link breakage problem and the high speed of vehicles have brought great challenges to the routing design of VANETs. It is difficult for existing routing protocols for WSN and MANETs to adapt the high dynamics of vehicular networks. Therefore, various number of routing protocols have been proposed to deal with the specific characteristics of VANETs. GPSR for Greedy Perimeter Stateless Routing is the most popular protocol, but it still suffers from frequent link breakages issue due to the high-mobility of vehicles, which cause a low PDR and throughput. In this investigation, we introduce two new strategies to enhance the classical GPSR protocol, by reducing the problem of link breakages and getting a stable route that improve the PDR and throughput in addition to reduce the routing overhead. The two proposed routing protocols E-GPSR and DVA-GPSR guide the selection of the next hop node based not only on the position but also on other important metrics of the participating nodes. Simulation results show that DVA-GPSR and E-GPSR protocols perform remarkably GPSR in terms of PDR, throughput, and routing overhead by varying the number of vehicles in a real urban scenario.","PeriodicalId":313743,"journal":{"name":"2020 International Conference on Intelligent Systems and Computer Vision (ISCV)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124478166","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 : 2020-06-01DOI: 10.1109/ISCV49265.2020.9204172
M. Chetto, Audrey Queudet
In many self-powered monitoring systems that support real-time applications, temporal guarantees are crucial. Energy-harvesting and energy-neutral sensing systems are such an example. In this paper, we survey selected prior work that addresses real-time issues in such systems. The survey outlines the requirements for optimal scheduling (i.e. clairvoyance and iding capabilities). The covered topics include fundamental results about EDF-based real-time scheduling, and real-time scheduling analysis of periodic task sets in energy harvesting systems. A detailed review is provided covering EDF and ED-H scheduling algorithms, their practicability, and the latest results from robustness and competitiveness investigations. The survey identifies key research challenges and likely productive research directions to help design and validation of any self-powered sensing system.
{"title":"EDF-based real-time scheduling for self-powered sensors: a survey of main theoretical results","authors":"M. Chetto, Audrey Queudet","doi":"10.1109/ISCV49265.2020.9204172","DOIUrl":"https://doi.org/10.1109/ISCV49265.2020.9204172","url":null,"abstract":"In many self-powered monitoring systems that support real-time applications, temporal guarantees are crucial. Energy-harvesting and energy-neutral sensing systems are such an example. In this paper, we survey selected prior work that addresses real-time issues in such systems. The survey outlines the requirements for optimal scheduling (i.e. clairvoyance and iding capabilities). The covered topics include fundamental results about EDF-based real-time scheduling, and real-time scheduling analysis of periodic task sets in energy harvesting systems. A detailed review is provided covering EDF and ED-H scheduling algorithms, their practicability, and the latest results from robustness and competitiveness investigations. The survey identifies key research challenges and likely productive research directions to help design and validation of any self-powered sensing system.","PeriodicalId":313743,"journal":{"name":"2020 International Conference on Intelligent Systems and Computer Vision (ISCV)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134094352","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 : 2020-06-01DOI: 10.1109/ISCV49265.2020.9204169
M. Yamni, H. Karmouni, M. Sayyouri, H. Qjidaa
This paper presents a new zero-watermarking scheme, for the copyright protection of stereo color images, based on the Quaternion Radial Tchebichef Moments (QRTMs) and on a chaotic system very sensitive to the initial values. The novelties of this scheme are as follows: (1) QRTMs are applied for the first time in the zero-watermarking scheme for stereo color image; (2) The QRTMs used in the proposed scheme have no numerical instability, which effectively improves the performance of the zero-watermarking scheme; (3) the proposed scheme based on QRTMs effectively resists geometric attacks such as: rotation, uniform scaling and non-uniform scaling; and (4) The chaotic system used is extremely dependent on the initial values, which offers the proposed scheme a high level of security. Experimental results show that the performance of the proposed scheme is superior to that of similar zero-watermarking schemes and other watermarking schemes of stereo color images.
{"title":"Color Stereo Image Zero-Watermarking using Quaternion Radial Tchebichef Moments","authors":"M. Yamni, H. Karmouni, M. Sayyouri, H. Qjidaa","doi":"10.1109/ISCV49265.2020.9204169","DOIUrl":"https://doi.org/10.1109/ISCV49265.2020.9204169","url":null,"abstract":"This paper presents a new zero-watermarking scheme, for the copyright protection of stereo color images, based on the Quaternion Radial Tchebichef Moments (QRTMs) and on a chaotic system very sensitive to the initial values. The novelties of this scheme are as follows: (1) QRTMs are applied for the first time in the zero-watermarking scheme for stereo color image; (2) The QRTMs used in the proposed scheme have no numerical instability, which effectively improves the performance of the zero-watermarking scheme; (3) the proposed scheme based on QRTMs effectively resists geometric attacks such as: rotation, uniform scaling and non-uniform scaling; and (4) The chaotic system used is extremely dependent on the initial values, which offers the proposed scheme a high level of security. Experimental results show that the performance of the proposed scheme is superior to that of similar zero-watermarking schemes and other watermarking schemes of stereo color images.","PeriodicalId":313743,"journal":{"name":"2020 International Conference on Intelligent Systems and Computer Vision (ISCV)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134185869","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 : 2020-06-01DOI: 10.1109/ISCV49265.2020.9204227
Hanene Mennour, S. Mostefai
This paper presents a hybrid deep learning neural network for classifying the network traffic data. In this regard, a Stacked Autoencoder and Feedforward neural network with tangent activation function have been employed. Firstly, we pre-trained the stacked Autoencoder with unsupervised learning method to improve the generalization of the classifier and limit the over-fitting problem in building the Feedforward neural network. in this state, the data is reconstructed into a new representation. After that, the Feedforward neural network as a supervised classifier has been stacked on the top. The purpose was to map the data in this new representation into class predictions. Finally, we have fine-tuned the entire network to accomplish the optimal hybrid model. A k fold cross-validation has been conducted to validate the system. CICIDS2017 datasets has been used in the experiment to classify normal and abnormal behaviour. The experimental results obtained by analyzing the proposed system show their superiority in terms of accuracy, detection rate and false alarm rate as compared to two state-of the-art machine learning algorithms which are Support Vector Machine (SVM) and Deep Neural Network. Our study achieves 98%, 100% for accuracy rates and F1 score respectively.
{"title":"A hybrid Deep Learning Strategy for an Anomaly Based N-IDS","authors":"Hanene Mennour, S. Mostefai","doi":"10.1109/ISCV49265.2020.9204227","DOIUrl":"https://doi.org/10.1109/ISCV49265.2020.9204227","url":null,"abstract":"This paper presents a hybrid deep learning neural network for classifying the network traffic data. In this regard, a Stacked Autoencoder and Feedforward neural network with tangent activation function have been employed. Firstly, we pre-trained the stacked Autoencoder with unsupervised learning method to improve the generalization of the classifier and limit the over-fitting problem in building the Feedforward neural network. in this state, the data is reconstructed into a new representation. After that, the Feedforward neural network as a supervised classifier has been stacked on the top. The purpose was to map the data in this new representation into class predictions. Finally, we have fine-tuned the entire network to accomplish the optimal hybrid model. A k fold cross-validation has been conducted to validate the system. CICIDS2017 datasets has been used in the experiment to classify normal and abnormal behaviour. The experimental results obtained by analyzing the proposed system show their superiority in terms of accuracy, detection rate and false alarm rate as compared to two state-of the-art machine learning algorithms which are Support Vector Machine (SVM) and Deep Neural Network. Our study achieves 98%, 100% for accuracy rates and F1 score respectively.","PeriodicalId":313743,"journal":{"name":"2020 International Conference on Intelligent Systems and Computer Vision (ISCV)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130914586","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：