Skip to main content

Alaa Tharwat

University of Applied Sciences Frankfurt, Computer Science and Engineering, Post-Doc

Followers

341

Following

84

Co-authors

7

Public Views

Alaa Tharwat received his BSc in 2002 from Mansoura University, MSc in 2008 Mansoura University, and PhD in 2017 from Suez Canal University.
He worked as an assistant lecturer at Gent University, within the framework
of the Welcome project – Erasmus Mundus Action 2 – with a title ‘Novel approach of multi-modal biometrics for animal identification’ in 2015.
Currently, he is an assistant lecturer in Faculty of Computer Science and Engineering, Frankfurt University of Applied Sciences, Frankfurt am Main, Germany. He has more than 30 scientific research papers published in prestigious international journals in the topics of machine learning and its applications.
Address: Germany- Frankfurt am Main

less

Masahiro Hattori

Abdelhameed Ibrahim

Mansoura University

Egyptian Nuclear Materials Authority.

Cairo University

John Tsiligaridis

University of North Carolina at Charlotte

Suganya Ramamoorthy

InterestsView All (8)

Uploads

Papers by Alaa Tharwat

Enzyme Function Classification

Handbook of Research on Machine Learning Innovations and Trends, 2017

Enzymes are important in our life and it plays a vital role in the most biological processes in t... more Enzymes are important in our life and it plays a vital role in the most biological processes in the living organisms and such as metabolic pathways. The classification of enzyme functionality from a sequence, structure data or the extracted features remains a challenging task. Traditional experiments consume more time, efforts, and cost. On the other hand, an automated classification of the enzymes saves efforts, money and time. The aim of this chapter is to cover and reviews the different approaches, which developed and conducted to classify and predict the functions of the enzyme proteins in addition to the new trends and challenges that could be considered now and in the future. The chapter addresses the main three approaches which are used in the classification the function of enzymatic proteins and illustrated the mechanism, pros, cons, and examples for each one.

A Novel Low-Query-Budget Active Learner with Pseudo-Labels for Imbalanced Data

Mathematics, 2022

Despite the availability of a large amount of free unlabeled data, collecting sufficient training... more Despite the availability of a large amount of free unlabeled data, collecting sufficient training data for supervised learning models is challenging due to the time and cost involved in the labeling process. The active learning technique we present here provides a solution by querying a small but highly informative set of unlabeled data. It ensures high generalizability across space, improving classification performance with test data that we have never seen before. Most active learners query either the most informative or the most representative data to annotate them. These two criteria are combined in the proposed algorithm by using two phases: exploration and exploitation phases. The former aims to explore the instance space by visiting new regions at each iteration. The second phase attempts to select highly informative points in uncertain regions. Without any predefined knowledge, such as initial training data, these two phases improve the search strategy of the proposed algori...

Personal Identification Based on Mobile-Based Keystroke Dynamics

Advances in Intelligent Systems and Computing, 2018

Ti t l e P e r s o n al id e n tific a tio n b a s e d o n m o bil e-b a s e d k ey s t r o k e d... more

How to Label? Combining Experts’ Knowledge for German Text Classification

2020 7th Swiss Conference on Data Science (SDS), 2020

A supervised machine learning classifier can only be as good as the labeled training data. For th... more A supervised machine learning classifier can only be as good as the labeled training data. For this reason, there is a need for explicit human expert knowledge inside the workflow. Existing data collections often consist of classes different to the ones which are necessary for an individual application. Therefore, generating a new data set based on a predefined labeling guideline is mandatory. The aim of this work is to increase the quality of labeled data sets during their creation. We present a workflow for the labeling of unsorted data by a group of experts, including subsequent classifier training and evaluation. Even if combined with standard methods for feature extraction and classification, a performance improvement was achieved with the proposed labeling method. Furthermore, we offer access to our data set (German newspaper articles) including the labeling guideline as contribution to the research community.

Population initialization techniques for evolutionary algorithms for single-objective constrained optimization problems: Deterministic vs. stochastic techniques

Swarm and Evolutionary Computation, 2021

Abstract The population initialization step is a common step in the majority (or even all) of evo... more Abstract The population initialization step is a common step in the majority (or even all) of evolutionary algorithms (EAs). There are many population initialization techniques. Due to the limited population size and the high dimensionality of many problems, there is little chance to cover the promising regions in the search space. From different perspectives, this paper compares the stochastic and deterministic population initialization techniques through comparing five of the well-known population initializers: Random number generator (RNG), Latin Hypercube, Sobol, Halton, and Kronecker. Due to the presence of many constraints in real-world applications, in this paper, we are focusing only on single-objective constrained optimization problems. Specifically, the goal is to investigate if there is a significant difference between these population initialization methods. In this paper, we explain theoretically and mathematically these different population initialization techniques. Moreover, different illustrative examples and visualizations are introduced to explain the behavior of each technique and compare different techniques from different perspectives. The results show that due to the high uniformity of the low-discrepancy sequences such as the Halton and Sobol sequences, the generated points using these sequences are more evenly distributed over the space than RNG, which is the commonly used technique for initializing the populations in EAs. Practically, using a set of benchmark functions, we investigate the use of each population initialization technique for initializing different population-based evolutionary algorithms. The results of our experiments prove that with sufficient numbers of iterations, the EAs are not sensitive to the initialization methods and there are no significant differences between the mentioned population initialization methods. Further, the low discrepancy methods enhance the exploration ability of EAs in early iterations.

Optimized superpixel and AdaBoost classifier for human thermal face recognition

Signal, Image and Video Processing, 2017

Chaotic dragonfly algorithm: an improved metaheuristic algorithm for feature selection

Applied Intelligence, 2018

Selecting the most discriminative features is a challenging problem in many applications. Bio-ins... more Selecting the most discriminative features is a challenging problem in many applications. Bio-inspired optimization algorithms have been widely applied to solve many optimization problems including the feature selection problem. In this paper, the most discriminating features were selected by a new Chaotic Dragonfly Algorithm (CDA) where chaotic maps embedded with searching iterations of the Dragonfly Algorithm (DA). Ten chaotic maps were employed to adjust the main parameters of dragonflies' movements through the optimization process to accelerate the convergence rate and improve the efficiency of DA. The proposed algorithm is employed for selecting features from the dataset that were extracted from the Drug bank database, which contained 6712 drugs. In this paper, 553 drugs that were bio-transformed into liver are used. This data have four toxic effects, namely, irritant, mutagenic, reproductive, and tumorigenic effect, where each drug is represented by 31 chemical descriptors. The proposed model is mainly comprised of three phases; data pre-processing, features selection, and the classification phase. In the data pre-processing phase, Synthetic Minority Over-sampling Technique (SMOTE) was used to solve the problem of the imbalanced dataset. At the features selection phase, the most discriminating features were selected using CDA. Finally, the selected features from CDA were used to feed Support Vector Machine (SVM) classifier at the classification phase. Experimental results proved the capability of CDA to find the optimal feature subset, which maximizing the classification performance and minimizing the number of selected features compared with DA and the other meta-heuristic optimization algorithms. Moreover, the experiments showed that Gauss chaotic map was the appropriate map to significantly boost the performance of DA. Additionally, the high obtained value of accuracy (81.82-96.08%), recall (80.84-96.11%), precision (81.45-96.08%) and F-Score (81.14-96.1%) for all toxic effects proved the robustness of the proposed model.

Classification assessment methods

Applied Computing and Informatics, 2018

Classification techniques have been applied to many applications in various fields of sciences. T... more Classification techniques have been applied to many applications in various fields of sciences. There are several ways of evaluating classification algorithms. The analysis of such metrics and its significance must be interpreted correctly for evaluating different learning algorithms. Most of these measures are scalar metrics and some of them are graphical methods. This paper introduces a detailed overview of the classification assessment measures with the aim of providing the basics of these measures and to show how it works to serve as a comprehensive source for researchers who are interested in this field. This overview starts by highlighting the definition of the confusion matrix in binary and multi-class classification problems. Many classification measures are also explained in details, and the influence of balanced and imbalanced data on each metric is presented. An illustrative example is introduced to show (1) how to calculate these measures in binary and multi-class classi...

A Wheelchair Control System Using Human-Machine Interaction: Single-Modal and Multimodal Approaches

Journal of Intelligent Systems, 2017

Recent research studies showed that brain-controlled systems/devices are breakthrough technology.... more Recent research studies showed that brain-controlled systems/devices are breakthrough technology. Such devices can provide disabled people with the power to control the movement of the wheelchair using different signals (e.g. EEG signals, head movements, and facial expressions). With this technology, disabled people can remotely steer a wheelchair, a computer, or a tablet. This paper introduces a simple, low-cost human-machine interface system to help chaired people to control their wheelchair using several control sources. To achieve this paper’s aim, a laptop was installed on a wheelchair in front of the sitting person, and the 14-electrode Emotiv EPOC headset was used to collect the person’s head impressions from the skull surface. The superficially picked-up signals, containing the brain thoughts, head gestures, and facial emotions, were electrically encoded and then wirelessly sent to a personal computer to be interpreted and then translated into useful control instructions. Us...

Bezier Curve Based Path Planning in a Dynamic Field using Modified Genetic Algorithm

Journal of Computational Science, 2017

Mobile robots have been used in different applications such as assembly, transportation, and manu... more Mobile robots have been used in different applications such as assembly, transportation, and manufacturing. Although, the great work to get the optimum robot's path, traditional path planning algorithms often assume that the environment is perfectly known and try to search for the optimal path that contains sharp turns and some polygonal lines. This paper proposes an efficient, Bezier curve based approach for the path planning in a dynamic field using a Modified Genetic Algorithm (MGA). The proposed MGA aims to boost the diversity of the generated solutions of the standard GA which increases the exploration capabilities of the MGA. In our proposed method, the robot's path is dynamically decided based on the obstacles' locations. With the goal of optimizing the distance between the start point and the target point, the MGA is employed to search for the most suitable points as the control points of the Bezier curve. Using the chosen control points, the optimum smooth path that minimizes the total distance between the start and the end points is selected. Our model was

Automatically Human Age Estimation Approach via Two-Dimensional Facial Image Analysis

Advances in Intelligent Systems and Computing, 2016

In this paper, the human age automatically estimated via two-dimensional facial image analysis. T... more In this paper, the human age automatically estimated via two-dimensional facial image analysis. The exact age estimation is often treated as a classification problem while it can be formulated as a regression problem. In our research, a classification and regression models are proposed. The two proposed models are evaluated using the same database images and the same features. Due to a big difference between the number of samples in each class or age group, the two proposed models used the complete and missing data in different experiments. Moreover, we compared age estimation errors when (1) Age estimation is performed without discrimination between males and females (gender unknown); (2) Age estimation is performed in males and females separately (gender known). Conclusions and results of these proposed models are shown by extensive experiments on the public available FG-NET database.

Moth-flame optimization for training Multi-Layer Perceptrons

2015 11th International Computer Engineering Conference (ICENCO), 2015

Multi-Layer Perceptron (MLP) is one of the Feed-Forward Neural Networks (FFNNs) types. Searching ... more Multi-Layer Perceptron (MLP) is one of the Feed-Forward Neural Networks (FFNNs) types. Searching for weights and biases in MLP is important to achieve minimum training error. In this paper, Moth-Flame Optimizer (MFO) is used to train Multi-Layer Perceptron (MLP). MFO-MLP is used to search for the weights and biases of the MLP to achieve minimum error and high classification rate. Five standard classification datasets are utilized to evaluate the performance of the proposed method. Moreover, three function-approximation datasets are used to test the performance of the proposed method. The proposed method (i.e. MFO-MLP) is compared with four well-known optimization algorithms, namely, Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Ant Colony Optimization (ACO), and Evolution Strategy (ES). The experimental results prove that the MFO algorithm is very competitive, solves the local optima problem, and it achieves a high accuracy.

2-Multimodal Biometric Authentication Algorithm Using Ear and Finger Knuckle Images ICCES 2012

Biometrics that use physiological traits such as face, iris, fingerprints, ear, and finger knuckl... more Biometrics that use physiological traits such as face, iris, fingerprints, ear, and finger knuckle (FK) for authentication face the problems of noisy sensors data, non-universality, and unacceptable error rates. Multimodal biometric methods use different fusion techniques to avoid such problems. Fusion methods have been proposed in different levels such as feature and classification level. This paper proposes two multimodal biometric authentication methods using ear and FK images. We propose a method based on fusion of images of ear and FK before the feature level, thus there is no information lost. We also propose a multi-level fusion method at image and classification levels. The features are extracted from the fused images using different classifiers and then combine the outputs of the classifiers in the abstract, rank, and score levels of fusion. Experimental results show that the proposed authentication methods increase the recognition rate compared to the state-of-the-art meth...

Balancing Exploration and Exploitation: A novel active learner for imbalanced data

Knowledge-Based Systems, 2020

Active learning receives great interest from researchers with the aim of reducing the amount of t... more Active learning receives great interest from researchers with the aim of reducing the amount of time, cost, and efforts for labeling data in many applications. Active learning aims to generate/select the smallest possible amount of training data that ensures strong classification performance in the test phase. An active learner carries out two main steps: (i) selecting a set of promising queries from unlabeled data, and (ii) annotating the selected queries. Most active learners choose either the most informative or representative instances for annotation. In this paper, we combined these two criteria for query selection. First, in the exploration phase, the proposed algorithm explores the search space and tries in each iteration to visit new regions for better exploration. This improves the capability of exploring the space of minority classes with imbalanced data. Second, in the exploitation phase, the goal is to generate a new point in an uncertain region, which is expected to be around the decision boundaries of the target functions. Some variants of the proposed algorithm do not require any labeled or unlabeled data in advance. There is only comparably few existing work which addresses this scenario. Experiments on synthetic and real datasets with different dimensions and imbalance ratios indicate that the proposed algorithm has significant advantages compared to various well-known active learners.

Classification error: Bias and variance, Underfitting and Overfitting

Particle Swarm Optimization

Handbook of Research on Machine Learning Innovations and Trends, 2017

Optimization algorithms are necessary to solve many problems such as parameter tuning. Particle S... more Optimization algorithms are necessary to solve many problems such as parameter tuning. Particle Swarm Optimization (PSO) is one of these optimization algorithms. The aim of PSO is to search for the optimal solution in the search space. This paper highlights the basic background needed to understand and implement the PSO algorithm. This paper starts with basic definitions of the PSO algorithm and how the particles are moved in the search space to find the optimal or near optimal solution. Moreover, a numerical example is illustrated to show how the particles are moved in a convex optimization problem. Another numerical example is illustrated to show how the PSO trapped in a local minima problem. Two experiments are conducted to show how the PSO searches for the optimal parameters in one-dimensional and two-dimensional spaces to solve machine learning problems.

A Predictive Model for Seminal Quality Using Neutrosophic Rule-Based Classification System

Advances in Intelligent Systems and Computing, 2018

This paper presents a predictive model of sperm quality based on personal lifestyle and environme... more This paper presents a predictive model of sperm quality based on personal lifestyle and environmental factors. Predicting the fertility based on measuring the quality of sperm is an urgent research problem since the fertility rates have been decreased mostly in men. Infertility is the biggest problem which faces the married persons. A lot of tests are expensive and time-consuming. Hence, these tests are not suitable for evaluating the quality of sperm; as a consequence, there is a need for implementing computational models that can predict the quality of sperm. The proposed model consists of two stages. In the first stage, Neutrosophic Rule-based Classification System (NRCS) and genetic NRCS (GNRCS) are proposed to classify an unknown seminal quality into normal or abnormal. NRCS and GNRCS forecast fertility based on nine input parameters which cover person lifestyle. The NRCS and GNRCS models were compared with three well-known classifiers such as decision trees, Support Vector Machines and Multilayer Perceptron with quality measures of GM, sensitivity, specificity, and accuracy. The proposed model (GNRCS) achieved promising results of 98.03% accuracy while basic NRCS achieved 95.55% accuracy. In the second stage, three different sampling algorithms are used to obtain balanced data.

Meta-Heuristic Algorithm Inspired by Grey Wolves for Solving Function Optimization Problems

In this paper, we suggest the use of Grey Wolf Optimization (GWO) algorithm to solve numerical op... more In this paper, we suggest the use of Grey Wolf Optimization (GWO) algorithm to solve numerical optimization problems. GWO is compared with two well-known optimization algorithms namely, Bat Algorithm (BA) and Particle Swarm Optimization (PSO), to test the improvement in the accuracy of finding the near optimal solution and the reduction in the computational cost. Ten standard benchmark functions were applied to test the performance of the three optimization algorithms in terms of accuracy and computational cost. The experimental results proved that our proposed method achieved accuracy better than the other two algorithms and it reduced the computational cost and converged rapidly to the optimal solution.

Combining evolutionary algorithms and case-based reasoning for learning high-quality shooting strategies in AI birds

2017 Evolving and Adaptive Intelligent Systems (EAIS), 2017

Self-adaptation and the ability to assimilate new knowledge are two fundamental characteristics o... more Self-adaptation and the ability to assimilate new knowledge are two fundamental characteristics of intelligent systems. In this paper we leverage methods from evolutionary optimization and from case-based reasoning to construct an agent that is able to evolve in such a way that it is able to successfully master the popular video game Angry Birds.

Communication in Soccer Simulation: On the Use of Wiretapping Opponent Teams

RoboCup 2018: Robot World Cup XXII, 2019

Inter-agent communication has been playing an important role in soccer simulation 2D since its in... more Inter-agent communication has been playing an important role in soccer simulation 2D since its introduction. Its primary usage has been to communicate with teammates in order to share state observations to fill gaps in the players’ world models, to announce near future actions like passes or requesting passes, as well as for sharing and synchronizing on locker room agreements. In this paper, by contrast, our focus is on the communication of the opponent team. We present an approach for wiretapping and decoding opponent communication and systematically evaluate its impact. Our main finding is that a team that wiretaps its opponent and exploits intercepted information appropriately, can boost its own playing performance significantly.

Enzyme Function Classification

Handbook of Research on Machine Learning Innovations and Trends, 2017

Enzymes are important in our life and it plays a vital role in the most biological processes in t... more Enzymes are important in our life and it plays a vital role in the most biological processes in the living organisms and such as metabolic pathways. The classification of enzyme functionality from a sequence, structure data or the extracted features remains a challenging task. Traditional experiments consume more time, efforts, and cost. On the other hand, an automated classification of the enzymes saves efforts, money and time. The aim of this chapter is to cover and reviews the different approaches, which developed and conducted to classify and predict the functions of the enzyme proteins in addition to the new trends and challenges that could be considered now and in the future. The chapter addresses the main three approaches which are used in the classification the function of enzymatic proteins and illustrated the mechanism, pros, cons, and examples for each one.

A Novel Low-Query-Budget Active Learner with Pseudo-Labels for Imbalanced Data

Mathematics, 2022

Despite the availability of a large amount of free unlabeled data, collecting sufficient training... more Despite the availability of a large amount of free unlabeled data, collecting sufficient training data for supervised learning models is challenging due to the time and cost involved in the labeling process. The active learning technique we present here provides a solution by querying a small but highly informative set of unlabeled data. It ensures high generalizability across space, improving classification performance with test data that we have never seen before. Most active learners query either the most informative or the most representative data to annotate them. These two criteria are combined in the proposed algorithm by using two phases: exploration and exploitation phases. The former aims to explore the instance space by visiting new regions at each iteration. The second phase attempts to select highly informative points in uncertain regions. Without any predefined knowledge, such as initial training data, these two phases improve the search strategy of the proposed algori...

Personal Identification Based on Mobile-Based Keystroke Dynamics

Advances in Intelligent Systems and Computing, 2018

Ti t l e P e r s o n al id e n tific a tio n b a s e d o n m o bil e-b a s e d k ey s t r o k e d... more

How to Label? Combining Experts’ Knowledge for German Text Classification

2020 7th Swiss Conference on Data Science (SDS), 2020

A supervised machine learning classifier can only be as good as the labeled training data. For th... more A supervised machine learning classifier can only be as good as the labeled training data. For this reason, there is a need for explicit human expert knowledge inside the workflow. Existing data collections often consist of classes different to the ones which are necessary for an individual application. Therefore, generating a new data set based on a predefined labeling guideline is mandatory. The aim of this work is to increase the quality of labeled data sets during their creation. We present a workflow for the labeling of unsorted data by a group of experts, including subsequent classifier training and evaluation. Even if combined with standard methods for feature extraction and classification, a performance improvement was achieved with the proposed labeling method. Furthermore, we offer access to our data set (German newspaper articles) including the labeling guideline as contribution to the research community.

Population initialization techniques for evolutionary algorithms for single-objective constrained optimization problems: Deterministic vs. stochastic techniques

Swarm and Evolutionary Computation, 2021

Abstract The population initialization step is a common step in the majority (or even all) of evo... more Abstract The population initialization step is a common step in the majority (or even all) of evolutionary algorithms (EAs). There are many population initialization techniques. Due to the limited population size and the high dimensionality of many problems, there is little chance to cover the promising regions in the search space. From different perspectives, this paper compares the stochastic and deterministic population initialization techniques through comparing five of the well-known population initializers: Random number generator (RNG), Latin Hypercube, Sobol, Halton, and Kronecker. Due to the presence of many constraints in real-world applications, in this paper, we are focusing only on single-objective constrained optimization problems. Specifically, the goal is to investigate if there is a significant difference between these population initialization methods. In this paper, we explain theoretically and mathematically these different population initialization techniques. Moreover, different illustrative examples and visualizations are introduced to explain the behavior of each technique and compare different techniques from different perspectives. The results show that due to the high uniformity of the low-discrepancy sequences such as the Halton and Sobol sequences, the generated points using these sequences are more evenly distributed over the space than RNG, which is the commonly used technique for initializing the populations in EAs. Practically, using a set of benchmark functions, we investigate the use of each population initialization technique for initializing different population-based evolutionary algorithms. The results of our experiments prove that with sufficient numbers of iterations, the EAs are not sensitive to the initialization methods and there are no significant differences between the mentioned population initialization methods. Further, the low discrepancy methods enhance the exploration ability of EAs in early iterations.

Optimized superpixel and AdaBoost classifier for human thermal face recognition

Signal, Image and Video Processing, 2017

Chaotic dragonfly algorithm: an improved metaheuristic algorithm for feature selection

Applied Intelligence, 2018

Selecting the most discriminative features is a challenging problem in many applications. Bio-ins... more Selecting the most discriminative features is a challenging problem in many applications. Bio-inspired optimization algorithms have been widely applied to solve many optimization problems including the feature selection problem. In this paper, the most discriminating features were selected by a new Chaotic Dragonfly Algorithm (CDA) where chaotic maps embedded with searching iterations of the Dragonfly Algorithm (DA). Ten chaotic maps were employed to adjust the main parameters of dragonflies' movements through the optimization process to accelerate the convergence rate and improve the efficiency of DA. The proposed algorithm is employed for selecting features from the dataset that were extracted from the Drug bank database, which contained 6712 drugs. In this paper, 553 drugs that were bio-transformed into liver are used. This data have four toxic effects, namely, irritant, mutagenic, reproductive, and tumorigenic effect, where each drug is represented by 31 chemical descriptors. The proposed model is mainly comprised of three phases; data pre-processing, features selection, and the classification phase. In the data pre-processing phase, Synthetic Minority Over-sampling Technique (SMOTE) was used to solve the problem of the imbalanced dataset. At the features selection phase, the most discriminating features were selected using CDA. Finally, the selected features from CDA were used to feed Support Vector Machine (SVM) classifier at the classification phase. Experimental results proved the capability of CDA to find the optimal feature subset, which maximizing the classification performance and minimizing the number of selected features compared with DA and the other meta-heuristic optimization algorithms. Moreover, the experiments showed that Gauss chaotic map was the appropriate map to significantly boost the performance of DA. Additionally, the high obtained value of accuracy (81.82-96.08%), recall (80.84-96.11%), precision (81.45-96.08%) and F-Score (81.14-96.1%) for all toxic effects proved the robustness of the proposed model.

Classification assessment methods

Applied Computing and Informatics, 2018

Classification techniques have been applied to many applications in various fields of sciences. T... more Classification techniques have been applied to many applications in various fields of sciences. There are several ways of evaluating classification algorithms. The analysis of such metrics and its significance must be interpreted correctly for evaluating different learning algorithms. Most of these measures are scalar metrics and some of them are graphical methods. This paper introduces a detailed overview of the classification assessment measures with the aim of providing the basics of these measures and to show how it works to serve as a comprehensive source for researchers who are interested in this field. This overview starts by highlighting the definition of the confusion matrix in binary and multi-class classification problems. Many classification measures are also explained in details, and the influence of balanced and imbalanced data on each metric is presented. An illustrative example is introduced to show (1) how to calculate these measures in binary and multi-class classi...

A Wheelchair Control System Using Human-Machine Interaction: Single-Modal and Multimodal Approaches

Journal of Intelligent Systems, 2017

Recent research studies showed that brain-controlled systems/devices are breakthrough technology.... more Recent research studies showed that brain-controlled systems/devices are breakthrough technology. Such devices can provide disabled people with the power to control the movement of the wheelchair using different signals (e.g. EEG signals, head movements, and facial expressions). With this technology, disabled people can remotely steer a wheelchair, a computer, or a tablet. This paper introduces a simple, low-cost human-machine interface system to help chaired people to control their wheelchair using several control sources. To achieve this paper’s aim, a laptop was installed on a wheelchair in front of the sitting person, and the 14-electrode Emotiv EPOC headset was used to collect the person’s head impressions from the skull surface. The superficially picked-up signals, containing the brain thoughts, head gestures, and facial emotions, were electrically encoded and then wirelessly sent to a personal computer to be interpreted and then translated into useful control instructions. Us...

Bezier Curve Based Path Planning in a Dynamic Field using Modified Genetic Algorithm

Journal of Computational Science, 2017

Mobile robots have been used in different applications such as assembly, transportation, and manu... more Mobile robots have been used in different applications such as assembly, transportation, and manufacturing. Although, the great work to get the optimum robot's path, traditional path planning algorithms often assume that the environment is perfectly known and try to search for the optimal path that contains sharp turns and some polygonal lines. This paper proposes an efficient, Bezier curve based approach for the path planning in a dynamic field using a Modified Genetic Algorithm (MGA). The proposed MGA aims to boost the diversity of the generated solutions of the standard GA which increases the exploration capabilities of the MGA. In our proposed method, the robot's path is dynamically decided based on the obstacles' locations. With the goal of optimizing the distance between the start point and the target point, the MGA is employed to search for the most suitable points as the control points of the Bezier curve. Using the chosen control points, the optimum smooth path that minimizes the total distance between the start and the end points is selected. Our model was

Automatically Human Age Estimation Approach via Two-Dimensional Facial Image Analysis

Advances in Intelligent Systems and Computing, 2016

In this paper, the human age automatically estimated via two-dimensional facial image analysis. T... more In this paper, the human age automatically estimated via two-dimensional facial image analysis. The exact age estimation is often treated as a classification problem while it can be formulated as a regression problem. In our research, a classification and regression models are proposed. The two proposed models are evaluated using the same database images and the same features. Due to a big difference between the number of samples in each class or age group, the two proposed models used the complete and missing data in different experiments. Moreover, we compared age estimation errors when (1) Age estimation is performed without discrimination between males and females (gender unknown); (2) Age estimation is performed in males and females separately (gender known). Conclusions and results of these proposed models are shown by extensive experiments on the public available FG-NET database.

Moth-flame optimization for training Multi-Layer Perceptrons

2015 11th International Computer Engineering Conference (ICENCO), 2015

Multi-Layer Perceptron (MLP) is one of the Feed-Forward Neural Networks (FFNNs) types. Searching ... more Multi-Layer Perceptron (MLP) is one of the Feed-Forward Neural Networks (FFNNs) types. Searching for weights and biases in MLP is important to achieve minimum training error. In this paper, Moth-Flame Optimizer (MFO) is used to train Multi-Layer Perceptron (MLP). MFO-MLP is used to search for the weights and biases of the MLP to achieve minimum error and high classification rate. Five standard classification datasets are utilized to evaluate the performance of the proposed method. Moreover, three function-approximation datasets are used to test the performance of the proposed method. The proposed method (i.e. MFO-MLP) is compared with four well-known optimization algorithms, namely, Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Ant Colony Optimization (ACO), and Evolution Strategy (ES). The experimental results prove that the MFO algorithm is very competitive, solves the local optima problem, and it achieves a high accuracy.

2-Multimodal Biometric Authentication Algorithm Using Ear and Finger Knuckle Images ICCES 2012

Biometrics that use physiological traits such as face, iris, fingerprints, ear, and finger knuckl... more Biometrics that use physiological traits such as face, iris, fingerprints, ear, and finger knuckle (FK) for authentication face the problems of noisy sensors data, non-universality, and unacceptable error rates. Multimodal biometric methods use different fusion techniques to avoid such problems. Fusion methods have been proposed in different levels such as feature and classification level. This paper proposes two multimodal biometric authentication methods using ear and FK images. We propose a method based on fusion of images of ear and FK before the feature level, thus there is no information lost. We also propose a multi-level fusion method at image and classification levels. The features are extracted from the fused images using different classifiers and then combine the outputs of the classifiers in the abstract, rank, and score levels of fusion. Experimental results show that the proposed authentication methods increase the recognition rate compared to the state-of-the-art meth...

Balancing Exploration and Exploitation: A novel active learner for imbalanced data

Knowledge-Based Systems, 2020

Active learning receives great interest from researchers with the aim of reducing the amount of t... more Active learning receives great interest from researchers with the aim of reducing the amount of time, cost, and efforts for labeling data in many applications. Active learning aims to generate/select the smallest possible amount of training data that ensures strong classification performance in the test phase. An active learner carries out two main steps: (i) selecting a set of promising queries from unlabeled data, and (ii) annotating the selected queries. Most active learners choose either the most informative or representative instances for annotation. In this paper, we combined these two criteria for query selection. First, in the exploration phase, the proposed algorithm explores the search space and tries in each iteration to visit new regions for better exploration. This improves the capability of exploring the space of minority classes with imbalanced data. Second, in the exploitation phase, the goal is to generate a new point in an uncertain region, which is expected to be around the decision boundaries of the target functions. Some variants of the proposed algorithm do not require any labeled or unlabeled data in advance. There is only comparably few existing work which addresses this scenario. Experiments on synthetic and real datasets with different dimensions and imbalance ratios indicate that the proposed algorithm has significant advantages compared to various well-known active learners.

Classification error: Bias and variance, Underfitting and Overfitting

Particle Swarm Optimization

Handbook of Research on Machine Learning Innovations and Trends, 2017

Optimization algorithms are necessary to solve many problems such as parameter tuning. Particle S... more Optimization algorithms are necessary to solve many problems such as parameter tuning. Particle Swarm Optimization (PSO) is one of these optimization algorithms. The aim of PSO is to search for the optimal solution in the search space. This paper highlights the basic background needed to understand and implement the PSO algorithm. This paper starts with basic definitions of the PSO algorithm and how the particles are moved in the search space to find the optimal or near optimal solution. Moreover, a numerical example is illustrated to show how the particles are moved in a convex optimization problem. Another numerical example is illustrated to show how the PSO trapped in a local minima problem. Two experiments are conducted to show how the PSO searches for the optimal parameters in one-dimensional and two-dimensional spaces to solve machine learning problems.

A Predictive Model for Seminal Quality Using Neutrosophic Rule-Based Classification System

Advances in Intelligent Systems and Computing, 2018

This paper presents a predictive model of sperm quality based on personal lifestyle and environme... more This paper presents a predictive model of sperm quality based on personal lifestyle and environmental factors. Predicting the fertility based on measuring the quality of sperm is an urgent research problem since the fertility rates have been decreased mostly in men. Infertility is the biggest problem which faces the married persons. A lot of tests are expensive and time-consuming. Hence, these tests are not suitable for evaluating the quality of sperm; as a consequence, there is a need for implementing computational models that can predict the quality of sperm. The proposed model consists of two stages. In the first stage, Neutrosophic Rule-based Classification System (NRCS) and genetic NRCS (GNRCS) are proposed to classify an unknown seminal quality into normal or abnormal. NRCS and GNRCS forecast fertility based on nine input parameters which cover person lifestyle. The NRCS and GNRCS models were compared with three well-known classifiers such as decision trees, Support Vector Machines and Multilayer Perceptron with quality measures of GM, sensitivity, specificity, and accuracy. The proposed model (GNRCS) achieved promising results of 98.03% accuracy while basic NRCS achieved 95.55% accuracy. In the second stage, three different sampling algorithms are used to obtain balanced data.

Meta-Heuristic Algorithm Inspired by Grey Wolves for Solving Function Optimization Problems

In this paper, we suggest the use of Grey Wolf Optimization (GWO) algorithm to solve numerical op... more In this paper, we suggest the use of Grey Wolf Optimization (GWO) algorithm to solve numerical optimization problems. GWO is compared with two well-known optimization algorithms namely, Bat Algorithm (BA) and Particle Swarm Optimization (PSO), to test the improvement in the accuracy of finding the near optimal solution and the reduction in the computational cost. Ten standard benchmark functions were applied to test the performance of the three optimization algorithms in terms of accuracy and computational cost. The experimental results proved that our proposed method achieved accuracy better than the other two algorithms and it reduced the computational cost and converged rapidly to the optimal solution.

Combining evolutionary algorithms and case-based reasoning for learning high-quality shooting strategies in AI birds

2017 Evolving and Adaptive Intelligent Systems (EAIS), 2017

Self-adaptation and the ability to assimilate new knowledge are two fundamental characteristics o... more Self-adaptation and the ability to assimilate new knowledge are two fundamental characteristics of intelligent systems. In this paper we leverage methods from evolutionary optimization and from case-based reasoning to construct an agent that is able to evolve in such a way that it is able to successfully master the popular video game Angry Birds.

Communication in Soccer Simulation: On the Use of Wiretapping Opponent Teams

RoboCup 2018: Robot World Cup XXII, 2019

Inter-agent communication has been playing an important role in soccer simulation 2D since its in... more Inter-agent communication has been playing an important role in soccer simulation 2D since its introduction. Its primary usage has been to communicate with teammates in order to share state observations to fill gaps in the players’ world models, to announce near future actions like passes or requesting passes, as well as for sharing and synchronizing on locker room agreements. In this paper, by contrast, our focus is on the communication of the opponent team. We present an approach for wiretapping and decoding opponent communication and systematically evaluate its impact. Our main finding is that a team that wiretaps its opponent and exploits intercepted information appropriately, can boost its own playing performance significantly.

Meta-heuristic Algorithms View project

This classifier (AdaBoost) is one of the important and efficient classifiers. It is one of the cl... more This classifier (AdaBoost) is one of the important and efficient classifiers. It is one of the classifier ensembles. This presentation presents the following:
• Explanation of the classifier ensemble concept.
• Explaining in details how the AdaBoost classifier works.
• Introduction how AdaBoost deals with
o Multiclassification
o Overfitting and underfitting
o Diversity
o Combining different outputs
• Comparing between Bagging and Boosting
• Presenting some application of the AdaBoost classifier.

Classification error: Bias and variance, Underfitting and Overfitting Python Code View project Using optimization algorithms to optimize SVM parameters View project

Bias and Variance errors