Pub Date : 2020-06-01DOI: 10.1109/ISCV49265.2020.9204225
Fatima Ezzahra Ezzrhari, Hassna Bensag, M. Youssfi, O. Bouattane, V. Kaburlasos
The field of multi agent systems (MAS) presents a multitude of middlewares allowing an ease to create and deploy applications of MAS. These middlewares are designed with programming models that strongly couple the communication framework of the agent and its cognitive pattern. Usually, more the number of agents used is large, more the communication model of the middleware is highly used and so the performance is impacted and perturbed.We present in this article a scalable multi-agent system middleware for High Performance Computing (HPC) of big data applications. Our proposed model is based on the principle of the separation between the learning pattern of the agent, its communication pattern and the data and processing distribution aspect. Our model is built around a set of layers based on APIs each having different implementations allowing the construction of agents, the communication of agents, the learning of agents, the distribution of data, the distribution of treatments, the construction and monitoring of the cluster.
{"title":"Scalable multi agent system middleware for HPC of Big Data Applications","authors":"Fatima Ezzahra Ezzrhari, Hassna Bensag, M. Youssfi, O. Bouattane, V. Kaburlasos","doi":"10.1109/ISCV49265.2020.9204225","DOIUrl":"https://doi.org/10.1109/ISCV49265.2020.9204225","url":null,"abstract":"The field of multi agent systems (MAS) presents a multitude of middlewares allowing an ease to create and deploy applications of MAS. These middlewares are designed with programming models that strongly couple the communication framework of the agent and its cognitive pattern. Usually, more the number of agents used is large, more the communication model of the middleware is highly used and so the performance is impacted and perturbed.We present in this article a scalable multi-agent system middleware for High Performance Computing (HPC) of big data applications. Our proposed model is based on the principle of the separation between the learning pattern of the agent, its communication pattern and the data and processing distribution aspect. Our model is built around a set of layers based on APIs each having different implementations allowing the construction of agents, the communication of agents, the learning of agents, the distribution of data, the distribution of treatments, the construction and monitoring of the cluster.","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":"117148794","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/ISCV49265.2020.9204218
Mohammed Qaraad, Souad Amjad, P. El-Kafrawy, Hanaa Fathi, Ibrahim I. M. Manhrawy
The feature selection method is regarded as an issue with the global combinatorial optimization technique, which aims to reduce the number of features, eliminate irrelevant, noisy and redundant data, such as microarray cancer data containing a small number of samples that have a large number of gene expression levels as features. To select the optimal subset of gene and reduce the dimensionality of cancer microarray data to improve the performance of the classification accuracy. This paper presents a model called PSO-ENSVM which is a hybrid between feature selection, optimization and classification methods. We use a Swarm optimization PSO algorithm which it's mainly the objective of this research is to have space to get near-optimal, optimal or solutions for optimizing the tuning parameters of Elastic Net and SVM as a classifier. To evaluate the model, we use seven microarray data sets for different cancer type, and we compared the PSO-ENSVM model with the PSO-SVM a model that optimizes RBF Kernel hyperparameter without feature selection and SVM with RBF Kernel. The experimental results were presented and showed that the ability of our model to obtain an ideal subset of the feature led to increased rates performance as it was able to reduce the number of features specified. As a result, the results show that the PSO-ENSVM model is superior compared to PSO-SVM and SVM with RBF kernel.
{"title":"Parameters Optimization of Elastic NET for High Dimensional Data using PSO Algorithm","authors":"Mohammed Qaraad, Souad Amjad, P. El-Kafrawy, Hanaa Fathi, Ibrahim I. M. Manhrawy","doi":"10.1109/ISCV49265.2020.9204218","DOIUrl":"https://doi.org/10.1109/ISCV49265.2020.9204218","url":null,"abstract":"The feature selection method is regarded as an issue with the global combinatorial optimization technique, which aims to reduce the number of features, eliminate irrelevant, noisy and redundant data, such as microarray cancer data containing a small number of samples that have a large number of gene expression levels as features. To select the optimal subset of gene and reduce the dimensionality of cancer microarray data to improve the performance of the classification accuracy. This paper presents a model called PSO-ENSVM which is a hybrid between feature selection, optimization and classification methods. We use a Swarm optimization PSO algorithm which it's mainly the objective of this research is to have space to get near-optimal, optimal or solutions for optimizing the tuning parameters of Elastic Net and SVM as a classifier. To evaluate the model, we use seven microarray data sets for different cancer type, and we compared the PSO-ENSVM model with the PSO-SVM a model that optimizes RBF Kernel hyperparameter without feature selection and SVM with RBF Kernel. The experimental results were presented and showed that the ability of our model to obtain an ideal subset of the feature led to increased rates performance as it was able to reduce the number of features specified. As a result, the results show that the PSO-ENSVM model is superior compared to PSO-SVM and SVM with RBF kernel.","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":"115328767","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/ISCV49265.2020.9204203
Sonia Souabi, A. Retbi, M. K. Idrissi, S. Bennani
Social networks are a powerful and efficient tool for e-learning promoting collaboration between learners. Thus, to better manage the learning process within these environments, it is imperative to use recommendation systems which take a very significant role in suggesting interesting material adapted to the different needs of learners. To model the recommendation systems, the researchers relied on numerous tools such as the exploitation of Machine Learning algorithms or social interactions between learners. Yet, behaviour within a social network can actually differ from one learner to another, so we will be dealing with several categories of learners with distinct attitudes. Based on this, we raise a rather important issue which is to classify the learners according to well-defined criteria and attitudes before calculating the recommendations. In the recommendation system we advocate, we therefore use the k-means algorithm to classify learners, then we calculate the recommendations for each cluster by referring to our old recommendation system proposed in one of our previous works. The global system is thus based on three essential points: k-means, correlation and co-occurrence. We then evaluate the performance of our proposed system in order to show its performance compared to the system that does not consider the k-means algorithm.
{"title":"A Recommendation Approach in Social Learning Based on K-Means Clustering","authors":"Sonia Souabi, A. Retbi, M. K. Idrissi, S. Bennani","doi":"10.1109/ISCV49265.2020.9204203","DOIUrl":"https://doi.org/10.1109/ISCV49265.2020.9204203","url":null,"abstract":"Social networks are a powerful and efficient tool for e-learning promoting collaboration between learners. Thus, to better manage the learning process within these environments, it is imperative to use recommendation systems which take a very significant role in suggesting interesting material adapted to the different needs of learners. To model the recommendation systems, the researchers relied on numerous tools such as the exploitation of Machine Learning algorithms or social interactions between learners. Yet, behaviour within a social network can actually differ from one learner to another, so we will be dealing with several categories of learners with distinct attitudes. Based on this, we raise a rather important issue which is to classify the learners according to well-defined criteria and attitudes before calculating the recommendations. In the recommendation system we advocate, we therefore use the k-means algorithm to classify learners, then we calculate the recommendations for each cluster by referring to our old recommendation system proposed in one of our previous works. The global system is thus based on three essential points: k-means, correlation and co-occurrence. We then evaluate the performance of our proposed system in order to show its performance compared to the system that does not consider the k-means algorithm.","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":"123480142","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/ISCV49265.2020.9204330
S. Sridhar, Siddartha Mootha, S. Subramanian
A stock market is a large trading environment, capable of handling millions of transactions. It is extremely difficult for regulatory bodies to manually detect whether a transaction was fraudulent or not. With the help of machine learning, it is possible to detect various scenarios of market manipulation. Market manipulation is when traders try to inflate or deflate the price of a stock to their advantage. This paper proposes to identify and detect market manipulation by implementing an Ensemble Neural Network. Our proposed system can identify three types of manipulation scenarios, i.e. Price manipulation, Volume Manipulation, and Trade Reversal. Based on the affidavit information provided by the Securities and Exchange Board of India (SEBI), a daily trading dataset was created from the Bombay Stock Exchange (BSE) website. The Ensemble Neural Network model with and without trainable sub-model layers was implemented on the daily trading dataset. The model with trainable sub-model layers achieved an accuracy of 91% and without trainable submodel layers achieved an accuracy of 96%
{"title":"Detection of Market Manipulation using Ensemble Neural Networks","authors":"S. Sridhar, Siddartha Mootha, S. Subramanian","doi":"10.1109/ISCV49265.2020.9204330","DOIUrl":"https://doi.org/10.1109/ISCV49265.2020.9204330","url":null,"abstract":"A stock market is a large trading environment, capable of handling millions of transactions. It is extremely difficult for regulatory bodies to manually detect whether a transaction was fraudulent or not. With the help of machine learning, it is possible to detect various scenarios of market manipulation. Market manipulation is when traders try to inflate or deflate the price of a stock to their advantage. This paper proposes to identify and detect market manipulation by implementing an Ensemble Neural Network. Our proposed system can identify three types of manipulation scenarios, i.e. Price manipulation, Volume Manipulation, and Trade Reversal. Based on the affidavit information provided by the Securities and Exchange Board of India (SEBI), a daily trading dataset was created from the Bombay Stock Exchange (BSE) website. The Ensemble Neural Network model with and without trainable sub-model layers was implemented on the daily trading dataset. The model with trainable sub-model layers achieved an accuracy of 91% and without trainable submodel layers achieved an accuracy of 96%","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":"122887925","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/ISCV49265.2020.9204325
Abdelilah Mbarek, Mouna Jiber, Ali Yahyaouy, Abdelouahed Sabri
Over the last decade, around 3500 people lost their lives in road accidents each year in Morocco. Between 2008 and 2017, the number of accidents has seen an increase of 38.11%. Several factors may contribute to the so-called “war on the roads”, such as the behavior of drivers or vehicle condition. Since human behavior is not always the leading cause of traffic crashes, in this work, we propose to study the effect of the environment and road conditions on accident mortality. The study is based on statistical data of accidents that caused death or bodily injuries in Morocco in 2017. The Case Fatality Rate (CFR) indicator was used to measure the severity of accidents, and the technique involved is the well-known non-parametric Analysis of Variance (ANOVA). Thirteen factors were taken into account to describe the state of the infrastructure and the physical conditions of roads. The analysis results show that the factors studied have a significant effect on accident fatality. More specifically, the type of intersection and the location proved to be the variables that contribute more to accident fatality.
{"title":"Road traffic mortality in Morocco: Analysis of statistical data","authors":"Abdelilah Mbarek, Mouna Jiber, Ali Yahyaouy, Abdelouahed Sabri","doi":"10.1109/ISCV49265.2020.9204325","DOIUrl":"https://doi.org/10.1109/ISCV49265.2020.9204325","url":null,"abstract":"Over the last decade, around 3500 people lost their lives in road accidents each year in Morocco. Between 2008 and 2017, the number of accidents has seen an increase of 38.11%. Several factors may contribute to the so-called “war on the roads”, such as the behavior of drivers or vehicle condition. Since human behavior is not always the leading cause of traffic crashes, in this work, we propose to study the effect of the environment and road conditions on accident mortality. The study is based on statistical data of accidents that caused death or bodily injuries in Morocco in 2017. The Case Fatality Rate (CFR) indicator was used to measure the severity of accidents, and the technique involved is the well-known non-parametric Analysis of Variance (ANOVA). Thirteen factors were taken into account to describe the state of the infrastructure and the physical conditions of roads. The analysis results show that the factors studied have a significant effect on accident fatality. More specifically, the type of intersection and the location proved to be the variables that contribute more to accident fatality.","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":"131359229","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/ISCV49265.2020.9204306
Hamed Laouz, Soheyb Ayad, L. Terrissa
Traffic accidents always cause great material and human losses. One of the most important causes of these accidents is the human factor, which is usually caused by fatigue or drowsiness. To address this problem, several approaches were proposed to predict the driver state. Some solutions are based on the measurement of the driver behavior such as: the head movement, the duration of the blink of the eye, the observation of the mouth expression. … etc., while the others are based on the measurements of the physiological signals to get information about the internal state of the driver’s body. These measurements are collected using different sensors such as Electrocardiogram (ECG), Electromyography (EMG), Electroencephalography (EEG), and Electrooculogram (EOG). In this paper, we presented a literature review on the recent related works in this field. In addition, we compared the methods used in each measurement approach. Finally, a detailed discussion according to the methods efficiency as well as the achieved results will be given.
{"title":"Literature Review on Driver’s Drowsiness and Fatigue Detection","authors":"Hamed Laouz, Soheyb Ayad, L. Terrissa","doi":"10.1109/ISCV49265.2020.9204306","DOIUrl":"https://doi.org/10.1109/ISCV49265.2020.9204306","url":null,"abstract":"Traffic accidents always cause great material and human losses. One of the most important causes of these accidents is the human factor, which is usually caused by fatigue or drowsiness. To address this problem, several approaches were proposed to predict the driver state. Some solutions are based on the measurement of the driver behavior such as: the head movement, the duration of the blink of the eye, the observation of the mouth expression. … etc., while the others are based on the measurements of the physiological signals to get information about the internal state of the driver’s body. These measurements are collected using different sensors such as Electrocardiogram (ECG), Electromyography (EMG), Electroencephalography (EEG), and Electrooculogram (EOG). In this paper, we presented a literature review on the recent related works in this field. In addition, we compared the methods used in each measurement approach. Finally, a detailed discussion according to the methods efficiency as well as the achieved results will be given.","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":"116940870","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/ISCV49265.2020.9204181
Karima Lamsaddak, D. Mentagui
Economic and social change in a digital era is making the insurance ecosystem more complex. It makes several agents interact for different purposes. Therefore, a reflection on the insurance ecosystem modelling through multi-agent simulation seems interesting since it allows to capture the complexity of today’s insured on the one hand and on the other hand to measure the solvency of insurance and reinsurance companies (EAR) in order to ensure the viability of the said ecosystem. Thus, this research aims at modeling the Moroccan insurance ecosystem within the framework of a new risk-based solvency directive for the case of loan death cover and on the basis of a set of exogenous and endogenous factors that influence the solvency of insurance and reinsurance companies in Morocco. This paper is carried out on the basis of a model, developed using NetLogo software, consisting of 4 agents that interact in the case of the “borrower’s death” guarantee, namely: the insured, the EAR, the banks and the Supervisory Authority of Insurance and Social Welfare (ACAPS). Each agent has a set of characteristics and seeks a defined objective. Thus, the modelling carried out allows testing the impact of endogenous and exogenous variables on the solvency of the EAR according to a simulation in three scenarios (central, rainy and risky).
{"title":"Multi-agent simulation of the Moroccan conventional insurance sector","authors":"Karima Lamsaddak, D. Mentagui","doi":"10.1109/ISCV49265.2020.9204181","DOIUrl":"https://doi.org/10.1109/ISCV49265.2020.9204181","url":null,"abstract":"Economic and social change in a digital era is making the insurance ecosystem more complex. It makes several agents interact for different purposes. Therefore, a reflection on the insurance ecosystem modelling through multi-agent simulation seems interesting since it allows to capture the complexity of today’s insured on the one hand and on the other hand to measure the solvency of insurance and reinsurance companies (EAR) in order to ensure the viability of the said ecosystem. Thus, this research aims at modeling the Moroccan insurance ecosystem within the framework of a new risk-based solvency directive for the case of loan death cover and on the basis of a set of exogenous and endogenous factors that influence the solvency of insurance and reinsurance companies in Morocco. This paper is carried out on the basis of a model, developed using NetLogo software, consisting of 4 agents that interact in the case of the “borrower’s death” guarantee, namely: the insured, the EAR, the banks and the Supervisory Authority of Insurance and Social Welfare (ACAPS). Each agent has a set of characteristics and seeks a defined objective. Thus, the modelling carried out allows testing the impact of endogenous and exogenous variables on the solvency of the EAR according to a simulation in three scenarios (central, rainy and risky).","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":"122649493","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/ISCV49265.2020.9204187
Fouad Nafis, Khalid Al Fararni, Ali Yahyaouy, Badraddine Aghoutane
In the Big data era, a large number of functionalities and applications are created. The immediate consequence is the loss of time for the user due to the difficulty of accessing relevant information, and therefore a questioning of the usefulness of the services offered. Recommender system (RS) aims to help potential users by recommending the most suitable offers according to their profiles and preferences, RSs based on collaborative filtering, or those based on content or even hybrid filtering have shown interesting results to be explored for the resolution of the problems encountered. But some limits remain unresolved which are mainly related to the ability of these techniques to build a robust and complete system capable of forming a complete idea of the user profile and then recommend them the most suitable offers. Hence, the advantage of using semantic RSs based on data web and semantic web technologies, specifically the ontologies. This paper offers a comparative study of existing semantic RSs in the field of cultural heritage in order to extract a complete vision of a RS for the scientific cultural heritage of the region of Drâa-Tafilalet in Morocco.
{"title":"Towards a semantic recommender system for cultural objects: Case study Draa-Tafilalet region","authors":"Fouad Nafis, Khalid Al Fararni, Ali Yahyaouy, Badraddine Aghoutane","doi":"10.1109/ISCV49265.2020.9204187","DOIUrl":"https://doi.org/10.1109/ISCV49265.2020.9204187","url":null,"abstract":"In the Big data era, a large number of functionalities and applications are created. The immediate consequence is the loss of time for the user due to the difficulty of accessing relevant information, and therefore a questioning of the usefulness of the services offered. Recommender system (RS) aims to help potential users by recommending the most suitable offers according to their profiles and preferences, RSs based on collaborative filtering, or those based on content or even hybrid filtering have shown interesting results to be explored for the resolution of the problems encountered. But some limits remain unresolved which are mainly related to the ability of these techniques to build a robust and complete system capable of forming a complete idea of the user profile and then recommend them the most suitable offers. Hence, the advantage of using semantic RSs based on data web and semantic web technologies, specifically the ontologies. This paper offers a comparative study of existing semantic RSs in the field of cultural heritage in order to extract a complete vision of a RS for the scientific cultural heritage of the region of Drâa-Tafilalet in Morocco.","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":"129610572","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/ISCV49265.2020.9204323
Kawtar Bouzoubaa, Youssef Taher, B. Nsiri
Today, individuals, business and public administrations are internet dependent. This strong dependence creates one of the important sources of threats. Among these threats, the famous Dos attack. The costs of downtime, outages and failures caused by these attacks are very important. Protecting and preventing these threats by using the conventional tools present important limits (cannot predict in real-time when, where, and how the new forms of these Dos attacks occur). To deal with these limits, cybersecurity systems based on machine learning models can analyze patterns and learn from them to forecast and prevent Dos attack. One of the key process which ensures the efficiency of these forecasting systems is feature selection. In this context, we paid particular attention to one of the efficient feature selection methods used in forecasting cybersecurity systems: Wrapper based-feature. To find the best subset of dos attack features and to optimize the accuracy of these systems, we present a comparative study between different wrapper methods applying to the dos attack forecasting. This investigation shows that a wrapper approach based on a genetic algorithm improves the forecasting accuracy more than other wrapper processes.
{"title":"Dos attack forecasting: A comparative study on wrapper feature selection","authors":"Kawtar Bouzoubaa, Youssef Taher, B. Nsiri","doi":"10.1109/ISCV49265.2020.9204323","DOIUrl":"https://doi.org/10.1109/ISCV49265.2020.9204323","url":null,"abstract":"Today, individuals, business and public administrations are internet dependent. This strong dependence creates one of the important sources of threats. Among these threats, the famous Dos attack. The costs of downtime, outages and failures caused by these attacks are very important. Protecting and preventing these threats by using the conventional tools present important limits (cannot predict in real-time when, where, and how the new forms of these Dos attacks occur). To deal with these limits, cybersecurity systems based on machine learning models can analyze patterns and learn from them to forecast and prevent Dos attack. One of the key process which ensures the efficiency of these forecasting systems is feature selection. In this context, we paid particular attention to one of the efficient feature selection methods used in forecasting cybersecurity systems: Wrapper based-feature. To find the best subset of dos attack features and to optimize the accuracy of these systems, we present a comparative study between different wrapper methods applying to the dos attack forecasting. This investigation shows that a wrapper approach based on a genetic algorithm improves the forecasting accuracy more than other wrapper processes.","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":"128556425","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/ISCV49265.2020.9204328
Safae Berrichi, A. Mazroui
The word alignment process, which is a critical step in statistical translation systems (SMT), has been suggested by several researchers as a promising track for enhancing neural translation system (NMT) performance in low-resource environments. Furthermore, given the negative impact on English/Arabic machine translation quality arising from the morphological richness and complexity of the Arabic language compared to the English language, we assessed in this study the relevance of the integration of morphosyntactic characteristics during the alignment phase. Indeed, we have enriched parallel corpora by morphosyntactic features such as stems, lemmas, roots, and POS tags; yet we have developed new SMT systems embedding one of these features in the word alignment phase. The test results proved the interest to use these features and highlighted the most relevant morphosyntactic information to the translation system.
{"title":"Enhancing Machine Translation by Integrating Linguistic Knowledge in the Word Alignment Module","authors":"Safae Berrichi, A. Mazroui","doi":"10.1109/ISCV49265.2020.9204328","DOIUrl":"https://doi.org/10.1109/ISCV49265.2020.9204328","url":null,"abstract":"The word alignment process, which is a critical step in statistical translation systems (SMT), has been suggested by several researchers as a promising track for enhancing neural translation system (NMT) performance in low-resource environments. Furthermore, given the negative impact on English/Arabic machine translation quality arising from the morphological richness and complexity of the Arabic language compared to the English language, we assessed in this study the relevance of the integration of morphosyntactic characteristics during the alignment phase. Indeed, we have enriched parallel corpora by morphosyntactic features such as stems, lemmas, roots, and POS tags; yet we have developed new SMT systems embedding one of these features in the word alignment phase. The test results proved the interest to use these features and highlighted the most relevant morphosyntactic information to the translation 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":"122600143","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：