Pub Date : 2020-10-24DOI: 10.1109/NILES50944.2020.9257935
M. H. Hassan, Omar A. Mahmoud, O. A. Mohammed, Ammar Y. Baraka, Amira T. Mahmoud, A. Yousef
Although many cross platform mobile development software used a trans-compiler-based approach, it was very difficult to generalize it to work in both directions. For example, to convert between Java for Android Development and Swift for iOS development and vice versa. This is due to the need of writing a specific parser for each source language, and a specific code generator for each destination language. Neural network-based models are used successfully to translate between natural languages, including English, French, German any many others by providing enough datasets and without the need of adding language specific code for understanding and generation. In this paper, a source code converter based on the Neural Machine Translation Transformer Model that can translate from Java to Swift and vice versa is introduced. A synthesized dataset is used to train the model, the pipeline used for the translation as well as the code synthesis procedure throughout the work are illustrated. Initial results are promising and give motivation to further enhance the proposed tool.
{"title":"Neural Machine Based Mobile Applications Code Translation","authors":"M. H. Hassan, Omar A. Mahmoud, O. A. Mohammed, Ammar Y. Baraka, Amira T. Mahmoud, A. Yousef","doi":"10.1109/NILES50944.2020.9257935","DOIUrl":"https://doi.org/10.1109/NILES50944.2020.9257935","url":null,"abstract":"Although many cross platform mobile development software used a trans-compiler-based approach, it was very difficult to generalize it to work in both directions. For example, to convert between Java for Android Development and Swift for iOS development and vice versa. This is due to the need of writing a specific parser for each source language, and a specific code generator for each destination language. Neural network-based models are used successfully to translate between natural languages, including English, French, German any many others by providing enough datasets and without the need of adding language specific code for understanding and generation. In this paper, a source code converter based on the Neural Machine Translation Transformer Model that can translate from Java to Swift and vice versa is introduced. A synthesized dataset is used to train the model, the pipeline used for the translation as well as the code synthesis procedure throughout the work are illustrated. Initial results are promising and give motivation to further enhance the proposed tool.","PeriodicalId":253090,"journal":{"name":"2020 2nd Novel Intelligent and Leading Emerging Sciences Conference (NILES)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2020-10-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123287386","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 : 2020-10-24DOI: 10.1109/NILES50944.2020.9257969
M. K. Diab, Ammar N. Abbas, H. Ammar, R. Shalaby
Detection of the lane boundary is the primary task in order to control the trajectory of an autonomous car. In this paper, three methodologies for lane detection are discussed with experimental illustration: Blob analysis, Hough transformation and Birds eye view. The next task after receiving the boundary points is to apply a control law in order to trigger the steering and velocity control to the motors efficiently. In the following, a comparative analysis is made between different tuning criteria to tune PID controller for Lane Keeping Assist (LKA). In order to receive the information of the environment a camera is used that sends wireless data to Simulink through Raspberry-Pi (R-Pi). The data is processed by the controller that transmits the desired output control to arduino through serial communication.
{"title":"Experimental Lane Keeping Assist for an Autonomous Vehicle Based on Optimal PID Controller","authors":"M. K. Diab, Ammar N. Abbas, H. Ammar, R. Shalaby","doi":"10.1109/NILES50944.2020.9257969","DOIUrl":"https://doi.org/10.1109/NILES50944.2020.9257969","url":null,"abstract":"Detection of the lane boundary is the primary task in order to control the trajectory of an autonomous car. In this paper, three methodologies for lane detection are discussed with experimental illustration: Blob analysis, Hough transformation and Birds eye view. The next task after receiving the boundary points is to apply a control law in order to trigger the steering and velocity control to the motors efficiently. In the following, a comparative analysis is made between different tuning criteria to tune PID controller for Lane Keeping Assist (LKA). In order to receive the information of the environment a camera is used that sends wireless data to Simulink through Raspberry-Pi (R-Pi). The data is processed by the controller that transmits the desired output control to arduino through serial communication.","PeriodicalId":253090,"journal":{"name":"2020 2nd Novel Intelligent and Leading Emerging Sciences Conference (NILES)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2020-10-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123396133","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 : 2020-10-24DOI: 10.1109/NILES50944.2020.9257977
Gad Gad, Ahmed Mahmoud Annaby, N. Negied, M. Darweesh
The rising interest in assistive and autonomous driving systems throughout the past decade has led to an active research community in perception and scene interpretation problems like lane detection. Traditional lane detection methods rely on specialized, hand-tailored features which is slow and prone to scalability. Recent methods that rely on deep learning and trained on pixel-wise lane segmentation have achieved better results and are able to generalize to a broad range of road and weather conditions. However, practical algorithms must be computationally inexpensive due to limited resources on vehicle-based platforms yet accurate to meet safety measures. In this approach, an encoder-decoder deep learning architecture generates binary segmentation of lanes, then the binary segmentation map is further processed to separate lanes, and a sliding window extracts each lane to produce the lane instance segmentation image. This method was validated on a tusimple data set, achieving competitive results.
{"title":"Real-Time Lane Instance Segmentation Using SegNet and Image Processing","authors":"Gad Gad, Ahmed Mahmoud Annaby, N. Negied, M. Darweesh","doi":"10.1109/NILES50944.2020.9257977","DOIUrl":"https://doi.org/10.1109/NILES50944.2020.9257977","url":null,"abstract":"The rising interest in assistive and autonomous driving systems throughout the past decade has led to an active research community in perception and scene interpretation problems like lane detection. Traditional lane detection methods rely on specialized, hand-tailored features which is slow and prone to scalability. Recent methods that rely on deep learning and trained on pixel-wise lane segmentation have achieved better results and are able to generalize to a broad range of road and weather conditions. However, practical algorithms must be computationally inexpensive due to limited resources on vehicle-based platforms yet accurate to meet safety measures. In this approach, an encoder-decoder deep learning architecture generates binary segmentation of lanes, then the binary segmentation map is further processed to separate lanes, and a sliding window extracts each lane to produce the lane instance segmentation image. This method was validated on a tusimple data set, achieving competitive results.","PeriodicalId":253090,"journal":{"name":"2020 2nd Novel Intelligent and Leading Emerging Sciences Conference (NILES)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2020-10-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115274040","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 : 2020-10-24DOI: 10.1109/NILES50944.2020.9257966
Ahmad El-Shiekh, Ahmad El-Alfy, A. Ammar, Mohamed Gamal, M. Dessouky, K. Salah, H. Mostafa
This paper proposes a new CAD tool that automates the RTL code generation based on the IPXACT standard (develop RTL code using XML files). Many related work generates RTL design using C language. In this work, the generation is based on XML descriptions. The tool is developed using Python. The generated RTL code can be synthesized by the synthesis tool like Design Compiler. Several commercial tools like MATLAB have this capability, but the proposed tool is faster and more configurable.
{"title":"IPXACT-Based RTL Generation Tool","authors":"Ahmad El-Shiekh, Ahmad El-Alfy, A. Ammar, Mohamed Gamal, M. Dessouky, K. Salah, H. Mostafa","doi":"10.1109/NILES50944.2020.9257966","DOIUrl":"https://doi.org/10.1109/NILES50944.2020.9257966","url":null,"abstract":"This paper proposes a new CAD tool that automates the RTL code generation based on the IPXACT standard (develop RTL code using XML files). Many related work generates RTL design using C language. In this work, the generation is based on XML descriptions. The tool is developed using Python. The generated RTL code can be synthesized by the synthesis tool like Design Compiler. Several commercial tools like MATLAB have this capability, but the proposed tool is faster and more configurable.","PeriodicalId":253090,"journal":{"name":"2020 2nd Novel Intelligent and Leading Emerging Sciences Conference (NILES)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2020-10-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133914508","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 : 2020-10-24DOI: 10.1109/NILES50944.2020.9257889
A. Hamdy, A. Ouda, A. Kamel, Y. Elhalwagy
Accurate positioning is required to achieve accurate navigation solution of moving objects therefore, inertial navigation system (INS) and global positioning system (GPS) are combined together to give a good solution in solving positioning problem and overcoming the problem of using each system separately. The motivation behind this work in this paper is to model and evaluate an INS/GPS integration algorithm model within 6-DOF flight simulation model by using loosely coupled integration technique and extended kalman filter (EKF) Algorithms to enhance and solve the position and attitude angles problems. Then, it is implemented on embedded microcontroller system (TM4C123GH6PM ARM Cortex-M4) using low-cost commercial sensors (MPU-6050 and GPS). Finally, the Navigation model is evaluated within 6-DOF simulation model using Processor-in-Loop (PIL) method. The system can realize comparable navigation accuracy with other high performance navigation system.
为了实现运动目标的精确导航，需要进行精确定位，因此将惯性导航系统(INS)和全球定位系统(GPS)结合起来，很好地解决了定位问题，克服了各自单独使用的问题。本文的工作动机是利用松耦合积分技术和扩展卡尔曼滤波(EKF)算法对六自由度飞行仿真模型中的INS/GPS集成算法模型进行建模和评估，以增强和解决位置和姿态角问题。然后，利用低成本的商用传感器(MPU-6050和GPS)在嵌入式微控制器系统(TM4C123GH6PM ARM Cortex-M4)上实现。最后，利用环内处理器(PIL)方法在六自由度仿真模型中对导航模型进行了评估。该系统可实现与其他高性能导航系统相当的导航精度。
{"title":"SoC loosely Coupled Navigation Algorithm Evaluation via 6-DOF Flight Simulation Model of Guided Bomb","authors":"A. Hamdy, A. Ouda, A. Kamel, Y. Elhalwagy","doi":"10.1109/NILES50944.2020.9257889","DOIUrl":"https://doi.org/10.1109/NILES50944.2020.9257889","url":null,"abstract":"Accurate positioning is required to achieve accurate navigation solution of moving objects therefore, inertial navigation system (INS) and global positioning system (GPS) are combined together to give a good solution in solving positioning problem and overcoming the problem of using each system separately. The motivation behind this work in this paper is to model and evaluate an INS/GPS integration algorithm model within 6-DOF flight simulation model by using loosely coupled integration technique and extended kalman filter (EKF) Algorithms to enhance and solve the position and attitude angles problems. Then, it is implemented on embedded microcontroller system (TM4C123GH6PM ARM Cortex-M4) using low-cost commercial sensors (MPU-6050 and GPS). Finally, the Navigation model is evaluated within 6-DOF simulation model using Processor-in-Loop (PIL) method. The system can realize comparable navigation accuracy with other high performance navigation system.","PeriodicalId":253090,"journal":{"name":"2020 2nd Novel Intelligent and Leading Emerging Sciences Conference (NILES)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2020-10-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114806352","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 : 2020-10-24DOI: 10.1109/NILES50944.2020.9257932
Ahmed N. Atiya, Hossam S. Hassan, Khaled E. Ibrahim, Omar M. ElGhandour, M. Tolba
The generation of Multi-scroll chaotic attractors and chaos theory has gained much attention due to its many usages in a wide range of applications such as image-encryption and random number generators. There have been many previous attempts to establish a system that is able to generate large numbers of n − scroll chaotic attractors by modifying existing systems such as Lorenz and Chua’s systems. In this paper, a proposed system based on generalizing Chua’s system that has shown its ability to produce an unprecedentedly large number of even and odd chaotic scrolls is introduced. MATLAB simulation is carried out to validate the proposed system and a GUI tool was developed to ease the process of generating any number of chaotic scrolls. Finally, an insight on how the proposed system can be generalized on the circuits level is given.
{"title":"Generalized Formula for Generating N-Scroll Chaotic Attractors","authors":"Ahmed N. Atiya, Hossam S. Hassan, Khaled E. Ibrahim, Omar M. ElGhandour, M. Tolba","doi":"10.1109/NILES50944.2020.9257932","DOIUrl":"https://doi.org/10.1109/NILES50944.2020.9257932","url":null,"abstract":"The generation of Multi-scroll chaotic attractors and chaos theory has gained much attention due to its many usages in a wide range of applications such as image-encryption and random number generators. There have been many previous attempts to establish a system that is able to generate large numbers of n − scroll chaotic attractors by modifying existing systems such as Lorenz and Chua’s systems. In this paper, a proposed system based on generalizing Chua’s system that has shown its ability to produce an unprecedentedly large number of even and odd chaotic scrolls is introduced. MATLAB simulation is carried out to validate the proposed system and a GUI tool was developed to ease the process of generating any number of chaotic scrolls. Finally, an insight on how the proposed system can be generalized on the circuits level is given.","PeriodicalId":253090,"journal":{"name":"2020 2nd Novel Intelligent and Leading Emerging Sciences Conference (NILES)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2020-10-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130594557","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 : 2020-10-24DOI: 10.1109/NILES50944.2020.9257951
Abdulmoneam A. Hassan, Laila H. Afify, A. El-Sherif, T. Elbatt
In this work, we aim at characterizing the aver-age success probability of content delivery in cache-equipped device-to-device (D2D) network under content-dependent channel access probability. We adopt retransmissions-upon-decoding-errors in a slotted-Aloha system, and account for the temporal interference correlation. We study the impact of the content-dependent access probabilities on the overall performance of the network. We verify the analytical results of this work via intensive Monte-Carlo simulations.
{"title":"Stochastic Modeling of Content-Dependent Scheduling in D2D Cache-Enabled Networks","authors":"Abdulmoneam A. Hassan, Laila H. Afify, A. El-Sherif, T. Elbatt","doi":"10.1109/NILES50944.2020.9257951","DOIUrl":"https://doi.org/10.1109/NILES50944.2020.9257951","url":null,"abstract":"In this work, we aim at characterizing the aver-age success probability of content delivery in cache-equipped device-to-device (D2D) network under content-dependent channel access probability. We adopt retransmissions-upon-decoding-errors in a slotted-Aloha system, and account for the temporal interference correlation. We study the impact of the content-dependent access probabilities on the overall performance of the network. We verify the analytical results of this work via intensive Monte-Carlo simulations.","PeriodicalId":253090,"journal":{"name":"2020 2nd Novel Intelligent and Leading Emerging Sciences Conference (NILES)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2020-10-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128823256","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 : 2020-10-24DOI: 10.1109/NILES50944.2020.9257903
Asmaa Elhadidy, Mohamed Afifi, Mohammed Hassoubah, Yara Ali, M. Elhelw
One of the key challenges in applying deep learning to solve real-life problems is the lack of large annotated datasets. Furthermore, for a deep learning model to perform well on the test set, all samples in the training and test sets should be independent and identically distributed (i.i.d.), which means that test samples should be similar to the samples that were used to train the model. In many cases, however, the underlying training and test set distributions are different. In such cases, it is common to adapt the test samples by transforming them to their equivalent counterparts in the domain of the training data before being processed by the deep learning model. In this paper, we perform domain adaptation of low-resolution 8, 16 and 32 channels LiDAR 3D point clouds projected on 2D spherical images in order to improve the quality of semantic segmentation tasks. To achieve this, the low-resolution 3D point clouds are transformed using an end-to-end supervised learning approach to spherical images that are very similar to those obtained by projecting high-resolution 64 channels LiDAR point clouds, without changing the underlying structure of the scene. The proposed framework is evaluated by training a semantic segmentation model on 64 channels LiDAR clouds from the Semantic KITTI dataset [1] and using this model to segment 8, 16 and 32 channel point clouds after adapting them using our framework. The results obtained from carried out experiments demonstrate the effectiveness of our framework where segmentation results surpassed those obtained with nearest neighbor interpolation methods.
{"title":"Improved Semantic Segmentation of Low-Resolution 3D Point Clouds Using Supervised Domain Adaptation","authors":"Asmaa Elhadidy, Mohamed Afifi, Mohammed Hassoubah, Yara Ali, M. Elhelw","doi":"10.1109/NILES50944.2020.9257903","DOIUrl":"https://doi.org/10.1109/NILES50944.2020.9257903","url":null,"abstract":"One of the key challenges in applying deep learning to solve real-life problems is the lack of large annotated datasets. Furthermore, for a deep learning model to perform well on the test set, all samples in the training and test sets should be independent and identically distributed (i.i.d.), which means that test samples should be similar to the samples that were used to train the model. In many cases, however, the underlying training and test set distributions are different. In such cases, it is common to adapt the test samples by transforming them to their equivalent counterparts in the domain of the training data before being processed by the deep learning model. In this paper, we perform domain adaptation of low-resolution 8, 16 and 32 channels LiDAR 3D point clouds projected on 2D spherical images in order to improve the quality of semantic segmentation tasks. To achieve this, the low-resolution 3D point clouds are transformed using an end-to-end supervised learning approach to spherical images that are very similar to those obtained by projecting high-resolution 64 channels LiDAR point clouds, without changing the underlying structure of the scene. The proposed framework is evaluated by training a semantic segmentation model on 64 channels LiDAR clouds from the Semantic KITTI dataset [1] and using this model to segment 8, 16 and 32 channel point clouds after adapting them using our framework. The results obtained from carried out experiments demonstrate the effectiveness of our framework where segmentation results surpassed those obtained with nearest neighbor interpolation methods.","PeriodicalId":253090,"journal":{"name":"2020 2nd Novel Intelligent and Leading Emerging Sciences Conference (NILES)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2020-10-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125970120","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 : 2020-10-24DOI: 10.1109/NILES50944.2020.9257883
A. Sobh, A. Kamel, A. Farouk, Y. Elhalwagy
This paper presents a design and evaluation for controlling a coupled system using a robust Linear Quadratic Regulator (LQR) controller acting on the augmented integral state space matrix model of a coupled system. System under investigation consisted of dual fan module that is interlinked and its axis moving freely in the pitch plan. On the other hand, a counter weight was used to balance the two fans thrust to optimize the controller effort in the elevation plan. The counterweight axe was denoted as the elevation axis. If the fans are not on the same horizontal line, the rotation of the system around itself in clockwise or anti-clockwise direction was carried out around the travel axis. The LQR controller design parameters should be able to stabilize itself at any degree on the travel or elevation axes while maintaining hover level along the pitch axis. Such controller acts by defining the penalty of each type of error in controlling this system. The error was multiplied by relevant penalty, then fed-back to the controller that controls the fan speeds accordingly. The representing model had three axes, each have a proportional, derivative, and integral term for the travel and elevation axes but not the pitch axis, the reasons will be discussed later in the paper. Modeling started by design process through defining a non-linear model of the system, linearizing it, then was transferred to state space format, add integral part to the model, then finally design and testing of an LQR controller.
{"title":"Three Dimension Angular Position Stabilization using LQR and Kalman Filter","authors":"A. Sobh, A. Kamel, A. Farouk, Y. Elhalwagy","doi":"10.1109/NILES50944.2020.9257883","DOIUrl":"https://doi.org/10.1109/NILES50944.2020.9257883","url":null,"abstract":"This paper presents a design and evaluation for controlling a coupled system using a robust Linear Quadratic Regulator (LQR) controller acting on the augmented integral state space matrix model of a coupled system. System under investigation consisted of dual fan module that is interlinked and its axis moving freely in the pitch plan. On the other hand, a counter weight was used to balance the two fans thrust to optimize the controller effort in the elevation plan. The counterweight axe was denoted as the elevation axis. If the fans are not on the same horizontal line, the rotation of the system around itself in clockwise or anti-clockwise direction was carried out around the travel axis. The LQR controller design parameters should be able to stabilize itself at any degree on the travel or elevation axes while maintaining hover level along the pitch axis. Such controller acts by defining the penalty of each type of error in controlling this system. The error was multiplied by relevant penalty, then fed-back to the controller that controls the fan speeds accordingly. The representing model had three axes, each have a proportional, derivative, and integral term for the travel and elevation axes but not the pitch axis, the reasons will be discussed later in the paper. Modeling started by design process through defining a non-linear model of the system, linearizing it, then was transferred to state space format, add integral part to the model, then finally design and testing of an LQR controller.","PeriodicalId":253090,"journal":{"name":"2020 2nd Novel Intelligent and Leading Emerging Sciences Conference (NILES)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2020-10-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125188621","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 : 2020-10-24DOI: 10.1109/NILES50944.2020.9257975
Abeer Youssef, M. Elattar, S. El-Beltagy
Named Entity Recognition (NER) is an important task in many natural language processing applications. There are several studies that have focused on NER for the English language. However, there are some limitations when applying the current methodologies directly on the Arabic language text. Recent studies have shown the effectiveness of pooled contextual embedding representations and significant improvements in English NER tasks. This work investigates the performance of pooled contextual embeddings and bidirectional encoder representations from Transformers (BERT) model when used for NER on the Arabic language while addressing Arabic specific issues. The proposed method is an end-to-end deep learning model that utilizes a combination of pre-trained word embeddings, pooled contextual embeddings, and BERT model. Embeddings are then fed into bidirectional long-short term memory networks with a conditional random field. Different types of classical and contextual embeddings were experimented to pool for the best model. The proposed method achieves an F1 score of 77.62% on the AQMAR dataset, outperforming all previously published results of deep learning, and non-deep learning models on the same dataset. The presented results also surpass those of the wining system for the same task on the same data in the Topcoder website competition.
{"title":"A Multi-Embeddings Approach Coupled with Deep Learning for Arabic Named Entity Recognition","authors":"Abeer Youssef, M. Elattar, S. El-Beltagy","doi":"10.1109/NILES50944.2020.9257975","DOIUrl":"https://doi.org/10.1109/NILES50944.2020.9257975","url":null,"abstract":"Named Entity Recognition (NER) is an important task in many natural language processing applications. There are several studies that have focused on NER for the English language. However, there are some limitations when applying the current methodologies directly on the Arabic language text. Recent studies have shown the effectiveness of pooled contextual embedding representations and significant improvements in English NER tasks. This work investigates the performance of pooled contextual embeddings and bidirectional encoder representations from Transformers (BERT) model when used for NER on the Arabic language while addressing Arabic specific issues. The proposed method is an end-to-end deep learning model that utilizes a combination of pre-trained word embeddings, pooled contextual embeddings, and BERT model. Embeddings are then fed into bidirectional long-short term memory networks with a conditional random field. Different types of classical and contextual embeddings were experimented to pool for the best model. The proposed method achieves an F1 score of 77.62% on the AQMAR dataset, outperforming all previously published results of deep learning, and non-deep learning models on the same dataset. The presented results also surpass those of the wining system for the same task on the same data in the Topcoder website competition.","PeriodicalId":253090,"journal":{"name":"2020 2nd Novel Intelligent and Leading Emerging Sciences Conference (NILES)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2020-10-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126722217","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：