Issam Bendib
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Issam Bendib has a total of 16 co-authors. Here are the top co-authors by number of shared publications:
Lamis
University of Larbi Tebessi,Tebessa, Algeria
2 shared publicationsMohamed Ridda Laouar
Laboratory of Mathematics, Informatics, and Systems (LAMIS), University of Larbi Tebessi, Tebessa, Algeria
2 shared publications- Mohamed Ridda Laouar
Echahid Cheikh Larbi Tebessi University
2 shared publications - Sarra Samet
Laboratory of Mathematics, Informatics, and Systems (LAMIS), University of Larbi Tebessi, Algeria
2 shared publications Zineb Touati-Hamad
Laboratory of Mathematics, Informatics, and Systems (LAMIS), University of Larbi Tebessi,Tebessa, Algeria
2 shared publications- Abdallah Meraoumia
University of Larbi Tebessi, Laboratory of Mathematics, Informatics and Systems (LAMIS), Tebessa
1 shared publication
Latest Publications
- Open Access
- 1 February 2026
- Published by Institute of Advanced Engineering and Science in Bulletin of Electrical Engineering and Informatics
- Vol. 15 (1) , 384-395
Facial expression recognition for emotional state identification using deep convolutional neural network
Facial expressions represent one of the most significant forms of non-verbal communication, with psychologists identifying six universal expressions: happiness, sadness, surprise, anger, fear, and disgust. Recognizing these expressions presents considerable challenges due to the subtlety of facial movements and variations across individuals. This paper presents a deep learning-based system for facial expression recognition (FER) that employs convolutional neural networks (CNNs) to classify emotional states. We investigate both a novel CNN architecture developed from scratch and established transfer learning approaches, evaluating their performance on the FER-2013 dataset. Our experimental results demonstrate that the proposed custom CNN architecture achieves 72.93% accuracy when combined with comprehensive data augmentation techniques, outperforming several baseline models. The system shows particular strength in recognizing fundamental emotions while maintaining computational efficiency suitable for real-time applications.
- 1 February 2026
- Published by Institute of Advanced Engineering and Science in Bulletin of Electrical Engineering and Informatics
- Vol. 15 (1) , 384-395
Open AccessFacial expression recognition for emotional state identification using deep convolutional neural network
Facial expressions represent one of the most significant forms of non-verbal communication, with psychologists identifying six universal expressions: happiness, sadness, surprise, anger, fear, and disgust. Recognizing these expressions presents considerable challenges due to the subtlety of facial movements and variations across individuals. This paper presents a deep learning-based system for facial expression recognition (FER) that employs convolutional neural networks (CNNs) to classify emotional states. We investigate both a novel CNN architecture developed from scratch and established transfer learning approaches, evaluating their performance on the FER-2013 dataset. Our experimental results demonstrate that the proposed custom CNN architecture achieves 72.93% accuracy when combined with comprehensive data augmentation techniques, outperforming several baseline models. The system shows particular strength in recognizing fundamental emotions while maintaining computational efficiency suitable for real-time applications.
- 1 July 2022
- Published by IGI Global Scientific Publishing in International Journal of Organizational and Collective Intelligence
- Vol. 12 (3) , 1-15
Towards Blockchain-Based Document Authentication
Nowadays, with the massive development of technologies, documents are usually made, modified and published digitally. Unlike paper documents, digital documents are easily falsified, and it is difficult to verify the authenticity of these advanced documents in effective and fast ways. Mushaf Al-Quran is one of the most sensitive digital documents that deserves to be preserved due to the divine messages it carries. There is no way to maintain an immutable document, as today's blockchain technology has proven. Therefore, to take advantage of blockchain technology, this article presents a decentralized web application (dApp) to authenticate and verify the digital Mushaf based on smart contracts and the Ethereum network. The results show the efficiency of the blockchain technology in preserving immutable documents and its effectiveness in protecting and proving the authenticity of the digital Mushaf Al-Quran.
- 1 July 2022
- Published by IGI Global Scientific Publishing in International Journal of Organizational and Collective Intelligence
- Vol. 12 (3) , 1-15
Towards Blockchain-Based Document Authentication
Nowadays, with the massive development of technologies, documents are usually made, modified and published digitally. Unlike paper documents, digital documents are easily falsified, and it is difficult to verify the authenticity of these advanced documents in effective and fast ways. Mushaf Al-Quran is one of the most sensitive digital documents that deserves to be preserved due to the divine messages it carries. There is no way to maintain an immutable document, as today's blockchain technology has proven. Therefore, to take advantage of blockchain technology, this article presents a decentralized web application (dApp) to authenticate and verify the digital Mushaf based on smart contracts and the Ethereum network. The results show the efficiency of the blockchain technology in preserving immutable documents and its effectiveness in protecting and proving the authenticity of the digital Mushaf Al-Quran.
- 1 May 2022
- Published by Bentham Science Publishers Ltd. in Recent Advances in Computer Science and Communications
- Vol. 15 (4) , 1-15
Improving Fine-grained Opinion Mining Approach with a Deep Constituency Tree-long Short Term Memory Network and Word Embedding
Objective: One key task of fine-grained opinion mining on product review is to extract product aspects and their corresponding opinion expressed by users. Previous work has demonstrated that precise modeling of opinion targets within the surrounding context can improve performances. However, how to effectively and efficiently learn hidden word semantics and better represent targets and the context still needs to be further studied. Recent years have seen a revival of the Long Short- Term Memory (LSTM), with its effectiveness being demonstrated on a wide range of problems. However, LSTM based approaches are still limited to linear data processing since it processes the information sequentially. As a result, they may perform poorly on user-generated texts, such as product reviews, tweets, etc., whose syntactic structure is not precise. Methods: In this research paper, we propose a constituency tree long short term memory neural network- based approach. We compare our model with state-of-the-art baselines on SemEval 2014 datasets. Results: Experiment results show that our models obtain competitive performances compared to various supervised LSTM architectures. Conclusion: Our work contributes to the improvement of state-of-the-art aspect-level opinion mining methods and offers a new approach to support the human decision-making process based on opinion mining results.
- 1 May 2022
- Published by Bentham Science Publishers Ltd. in Recent Advances in Computer Science and Communications
- Vol. 15 (4) , 1-15
Improving Fine-grained Opinion Mining Approach with a Deep Constituency Tree-long Short Term Memory Network and Word Embedding
Objective: One key task of fine-grained opinion mining on product review is to extract product aspects and their corresponding opinion expressed by users. Previous work has demonstrated that precise modeling of opinion targets within the surrounding context can improve performances. However, how to effectively and efficiently learn hidden word semantics and better represent targets and the context still needs to be further studied. Recent years have seen a revival of the Long Short- Term Memory (LSTM), with its effectiveness being demonstrated on a wide range of problems. However, LSTM based approaches are still limited to linear data processing since it processes the information sequentially. As a result, they may perform poorly on user-generated texts, such as product reviews, tweets, etc., whose syntactic structure is not precise. Methods: In this research paper, we propose a constituency tree long short term memory neural network- based approach. We compare our model with state-of-the-art baselines on SemEval 2014 datasets. Results: Experiment results show that our models obtain competitive performances compared to various supervised LSTM architectures. Conclusion: Our work contributes to the improvement of state-of-the-art aspect-level opinion mining methods and offers a new approach to support the human decision-making process based on opinion mining results.
- 1 January 2022
- Published by IGI Global Scientific Publishing in International Journal of Decision Support System Technology
- Vol. 14 (1) , 1-19
Analysis and Prediction of Diabetes Disease Using Machine Learning Methods
To increase healthcare quality, early illness prediction helps patients prevent potentially life-threatening health issues before it is too late. Artificial intelligence is a rapidly evolving area, and its applications to diabetes, a worldwide epidemic, have the potential to revolutionize the way diabetes is diagnosed and managed. A total of six supervised machine learning algorithms based on patient data were used and compared to predict the diagnosis of diabetes mellitus. For experiments, the Pima Indians Diabetes Database was used, and their missing values were carefully handled by different techniques. For random train-test splits, the Random Forest classification algorithm achieved an accuracy rate of 92 percent. This model outperforms other state-of-the-art approaches due to the application of a combination of techniques for dealing with missing values (the mixture of imputing missing values techniques) that is proposed. With this approach, the models of this manuscript achieved better accuracy than prior work done with the Pima diabetes data.
- 1 January 2022
- Published by IGI Global Scientific Publishing in International Journal of Decision Support System Technology
- Vol. 14 (1) , 1-19
Analysis and Prediction of Diabetes Disease Using Machine Learning Methods
To increase healthcare quality, early illness prediction helps patients prevent potentially life-threatening health issues before it is too late. Artificial intelligence is a rapidly evolving area, and its applications to diabetes, a worldwide epidemic, have the potential to revolutionize the way diabetes is diagnosed and managed. A total of six supervised machine learning algorithms based on patient data were used and compared to predict the diagnosis of diabetes mellitus. For experiments, the Pima Indians Diabetes Database was used, and their missing values were carefully handled by different techniques. For random train-test splits, the Random Forest classification algorithm achieved an accuracy rate of 92 percent. This model outperforms other state-of-the-art approaches due to the application of a combination of techniques for dealing with missing values (the mixture of imputing missing values techniques) that is proposed. With this approach, the models of this manuscript achieved better accuracy than prior work done with the Pima diabetes data.
- 27 December 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-8
Predicting and Staging Chronic Kidney Disease using Optimized Random Forest Algorithm
The silent killer Chronic Kidney Disease (CKD) in wealthy countries and listed with the leading causes of death in impoverished countries. Because of its rising incidence, CKD is included in the most serious public health problems. It is apparent that early detection of CKD may reduce the severity of damage in maturity. The patient must go to a diagnostic facility and consult with a doctor. This significant issue has been solved with the introduction of machine learning. This study’s main objective is to build a model that can reliably predict a person’s risk of acquiring CKD. Data mining and machine learning techniques have been widely employed for forecasting chronic renal disease, but little research has been done mixing imputation approaches at the pre-processing stage and feature selection strategy so that classification accuracy will be enhanced. The CKD Database, which is used in the experiments and consists of 400 records with 25, is accessible through UCI’s machine learning repository. It does, however, have a large number of missing values, which is why we proposed combining several missing data imputation strategies to solve the problem. The chi-square test was used to select features in this work. A supervised machine learning classification model called Random Forest (RF) is utilized and optimized with gridsearch to diagnose CKD at an early stage. Following a cross-validation procedure with 5 folders, several metrics were utilized to evaluate the model. Our RF had a 99.24% accuracy. The model’s best result is created by considering the 10 best-selected features. When compared to previous studies, our results are among the best for assessment metrics and the ranking accuracy. However, with only fewer features. In practice, some decision assistance for renal illness’ diagnosis, prevention, and prediction are provided by this study.
- 27 December 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-8
Predicting and Staging Chronic Kidney Disease using Optimized Random Forest Algorithm
The silent killer Chronic Kidney Disease (CKD) in wealthy countries and listed with the leading causes of death in impoverished countries. Because of its rising incidence, CKD is included in the most serious public health problems. It is apparent that early detection of CKD may reduce the severity of damage in maturity. The patient must go to a diagnostic facility and consult with a doctor. This significant issue has been solved with the introduction of machine learning. This study’s main objective is to build a model that can reliably predict a person’s risk of acquiring CKD. Data mining and machine learning techniques have been widely employed for forecasting chronic renal disease, but little research has been done mixing imputation approaches at the pre-processing stage and feature selection strategy so that classification accuracy will be enhanced. The CKD Database, which is used in the experiments and consists of 400 records with 25, is accessible through UCI’s machine learning repository. It does, however, have a large number of missing values, which is why we proposed combining several missing data imputation strategies to solve the problem. The chi-square test was used to select features in this work. A supervised machine learning classification model called Random Forest (RF) is utilized and optimized with gridsearch to diagnose CKD at an early stage. Following a cross-validation procedure with 5 folders, several metrics were utilized to evaluate the model. Our RF had a 99.24% accuracy. The model’s best result is created by considering the 10 best-selected features. When compared to previous studies, our results are among the best for assessment metrics and the ranking accuracy. However, with only fewer features. In practice, some decision assistance for renal illness’ diagnosis, prevention, and prediction are provided by this study.
- Open Access
- 1 October 2021
- Published by IGI Global Scientific Publishing in International Journal of Healthcare Information Systems and Informatics
- Vol. 16 (4) , 1-23
Transfer Learning for Highlighting Diagnosis in Pathological Anatomy Based on Immunohistochemistry
In the medical field, the diagnostic phase is the most important, as the entire treatment process will be based on this step. Oncological diseases such as breast cancer require a precise anatomopathological study accompanied most of the time by an immunohistochemical study whose goal is to know the sensitivity of tumor tissues to hormone therapy and targeted therapy. This study relies on antibodies and their interpretation requires significant time and as it can suffer from poor reproducibility which negatively influences the treatment stage. In this work, the objective is to classify histopathological images stained with E-cadherin antibody to help pathologists in their work in order to facilitate oncologists in the choice of the most appropriate therapeutic protocol. The realization of this task is based on the choice of transfer learning as techniques and data augmentation due to the minimal number of images gathered. The results obtained are very satisfying both on accuracy where we reached a rate of 97.27% with a reduced number of parameters and very close to our basic model.
- 1 October 2021
- Published by IGI Global Scientific Publishing in International Journal of Healthcare Information Systems and Informatics
- Vol. 16 (4) , 1-23
Open AccessTransfer Learning for Highlighting Diagnosis in Pathological Anatomy Based on Immunohistochemistry
In the medical field, the diagnostic phase is the most important, as the entire treatment process will be based on this step. Oncological diseases such as breast cancer require a precise anatomopathological study accompanied most of the time by an immunohistochemical study whose goal is to know the sensitivity of tumor tissues to hormone therapy and targeted therapy. This study relies on antibodies and their interpretation requires significant time and as it can suffer from poor reproducibility which negatively influences the treatment stage. In this work, the objective is to classify histopathological images stained with E-cadherin antibody to help pathologists in their work in order to facilitate oncologists in the choice of the most appropriate therapeutic protocol. The realization of this task is based on the choice of transfer learning as techniques and data augmentation due to the minimal number of images gathered. The results obtained are very satisfying both on accuracy where we reached a rate of 97.27% with a reduced number of parameters and very close to our basic model.
- 21 September 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 25, 1-4
Authentication of Quran Verses Sequences Using Deep Learning
The emergence of electronic copies of the Holy Quran has allowed an increase in the phenomenon of reading and sharing Arabic verses among internet users. The Quran has been brought together in one book to maintain the order and integrity of its content. Changing the order of verses is considered as a degradation of the Holy Quran, resulting in new surahs in the form of a new composition, contradicting the structure and meaning of what is in the Ottoman Quran. Similar to any verification methodology, this study aims to apply deep learning algorithms to automatically authenticate the integrity of the Quranic content arrangement. The Long Short-Term Memory (LSTM) algorithm was chosen in this work, and the results achieved a test accuracy of 99.98% on the dataset that we created using Tanzil website data.
- 21 September 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 25, 1-4
Authentication of Quran Verses Sequences Using Deep Learning
The emergence of electronic copies of the Holy Quran has allowed an increase in the phenomenon of reading and sharing Arabic verses among internet users. The Quran has been brought together in one book to maintain the order and integrity of its content. Changing the order of verses is considered as a degradation of the Holy Quran, resulting in new surahs in the form of a new composition, contradicting the structure and meaning of what is in the Ottoman Quran. Similar to any verification methodology, this study aims to apply deep learning algorithms to automatically authenticate the integrity of the Quranic content arrangement. The Long Short-Term Memory (LSTM) algorithm was chosen in this work, and the results achieved a test accuracy of 99.98% on the dataset that we created using Tanzil website data.
- 1 May 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 229-234
Does the Learning Principle Help Us Improve Hand-Crafted Finger-Knuckle-Print Features?
One of the most critical steps in pattern recognition systems is to extract the most distinctive features of the image. Therefore, the accuracy of these systems is strongly related to the efficiency of this step to represent the image. In general, hand-craft image features are obtained using feature extraction methods that often have no prior knowledge about the image type, which may result in poor features. In this paper, a hand-craft image feature learning method, based on Local Binary Pattern (LBP), for Finger-Knuckle-Print representation and recognition is applied. Thus, in the learning phase, each training finger image is first divided into several non-overlapping regions, and the feature filter and the codebook of each region are individually learned. For the enrolment phase, the obtained learned filter and codebook are applied to each image to extract the histogram feature of each region. Then, the different histograms are concatenated in a single feature vector to represent the finger features. Finally, an SVM classifier is used for the identification phase. The experimental results clearly show that the learning principle offers a good gain in the system accuracy, which can provide an excellent identification rate and thus improve security.
- 1 May 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 229-234
Does the Learning Principle Help Us Improve Hand-Crafted Finger-Knuckle-Print Features?
One of the most critical steps in pattern recognition systems is to extract the most distinctive features of the image. Therefore, the accuracy of these systems is strongly related to the efficiency of this step to represent the image. In general, hand-craft image features are obtained using feature extraction methods that often have no prior knowledge about the image type, which may result in poor features. In this paper, a hand-craft image feature learning method, based on Local Binary Pattern (LBP), for Finger-Knuckle-Print representation and recognition is applied. Thus, in the learning phase, each training finger image is first divided into several non-overlapping regions, and the feature filter and the codebook of each region are individually learned. For the enrolment phase, the obtained learned filter and codebook are applied to each image to extract the histogram feature of each region. Then, the different histograms are concatenated in a single feature vector to represent the finger features. Finally, an SVM classifier is used for the identification phase. The experimental results clearly show that the learning principle offers a good gain in the system accuracy, which can provide an excellent identification rate and thus improve security.