Claim Missing Document
Check
Articles

Found 8 Documents
Search

Optimizing credit card fraud detection: a deep learning approach to imbalanced datasets Ndama, Oussama; Bensassi, Ismail; En-Naimi, El Mokhtar
International Journal of Electrical and Computer Engineering (IJECE) Vol 14, No 4: August 2024
Publisher : Institute of Advanced Engineering and Science

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.11591/ijece.v14i4.pp4802-4814

Abstract

Imbalanced datasets pose a significant challenge in credit card fraud detection, hindering the training effectiveness of models due to the scarcity of fraudulent cases. This study addresses the critical problem of data imbalance through an in-depth exploration of techniques, including cross-entropy loss minimization, weighted optimization, and synthetic minority oversampling technique-based resampling, coupled with deep neural networks (DNNs). The urgent need to combat class imbalances in credit card fraud datasets is underscored, emphasizing the creation of reliable detection models. The research method delves into the application of DNNs, strategically optimizing and resampling the dataset to enhance model performance. The study employs a dataset from October 2018, containing 284,807 transactions, with a mere 492 classified as fraudulent. Various resampling techniques, such as undersampling and SMOTE oversampling, are evaluated alongside weighted optimization. The results showcase the effectiveness of SMOTE oversampling, achieving an accuracy of 99.83% without any false negatives. The study concludes by advocating for flexible strategies, integrating cutting-edge machine learning methods, and developing adaptive defenses to safeguard against emerging financial risks in credit card fraud detection.
Innovative credit card fraud detection: A hybrid model combining artificial neural networks and support vector machines Ndama, Oussama; Bensassi, Ismail; En-naimi, El Mokhtar
IAES International Journal of Artificial Intelligence (IJ-AI) Vol 13, No 3: September 2024
Publisher : Institute of Advanced Engineering and Science

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.11591/ijai.v13.i3.pp2674-2682

Abstract

In recent years, escalating fraudulent activities have led to significant financial losses across industries, intensifying the critical challenge of fraud detection. This study introduces a novel hybrid model that combines artificial neural networks (ANN) with support vector machines (SVM) to construct a robust additive model for fraud detection. Emphasizing the Synthetic Minority Over-sampling Technique (SMOTE), our investigation addresses the imbalanced nature of fraud versus non-fraud transactions. The clear novelty of our research lies in the seamless integration of these two powerful tools, offering a comprehensive and effective solution to the challenges posed by credit card fraud detection. Furthermore, our study stands out by emphasizing the collaborative synergy between ANN and SVM, particularly through the integration of multiple kernels, which improves the adaptability and accuracy of the proposed hybrid model. We conducted a thorough examination of 284,807 anonymized transactions, placing special emphasis on comparing the hybrid approach's performance and showcasing its superiority over traditional methodologies in the realm of fraud detection.
Hybrid RNNs and USE for enhanced sequential sentence classification in biomedical paper abstracts Ndama, Oussama; Bensassi, Ismail; En-Naimi, El Mokhtar
Bulletin of Electrical Engineering and Informatics Vol 13, No 5: October 2024
Publisher : Institute of Advanced Engineering and Science

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.11591/eei.v13i5.8240

Abstract

This research evaluates a number of hybrid recurrent neural network (RNN) architectures for classifying sequential sentences in biomedical abstracts. The architectures include long short-term memory (LSTM), bidirectional LSTM (BI-LSTM), gated recurrent unit (GRU), and bidirectional GRU (BI-GRU) models, all of which are combined with the universal sentence encoder (USE). The investigation assesses their efficacy in categorizing sentences into predefined classes: background, objective, method, result, and conclusion. Each RNN variant is used with the pre-trained USE as word embeddings to find complex sequential relationships in biomedical text. Results demonstrate the adaptability and effectiveness of these hybrid architectures in discerning diverse sentence functions. This research addresses the need for improved literature comprehension in biomedicine by employing automated sentence classification techniques, highlighting the significance of advanced hybrid algorithms in enhancing text classification methodologies within biomedical research.
Enhancing fake profile detection through supervised and hybrid machine learning: a comparative analysis Bensassi, Ismail; Ndama, Oussama; Kouissi, Mohamed; En-Naimi, El Mokhtar
Indonesian Journal of Electrical Engineering and Computer Science Vol 37, No 1: January 2025
Publisher : Institute of Advanced Engineering and Science

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.11591/ijeecs.v37.i1.pp257-268

Abstract

In modern times, social networks have become ubiquitous platforms facilitating widespread information dissemination, resulting in significant daily data generation. This increase in data production encompasses a wide range of user-generated content, which in turn promotes the proliferation of fraudulent users creating fake profiles and engaging in deceptive activities. This article aims to address this challenge by employing machine learning algorithms to accurately identify fake profiles. The research involves a thorough analysis of various user behaviors, engagement metrics, and content attributes within social platforms. The primary goal is to develop robust models capable of effectively detecting deceptive profiles by meticulously examining user activities and content characteristics. The study explores the application of robust methodologies such as K-means and K-medoids clustering, alongside supervised machine learning classifiers including K-nearest neighbors (KNN), support vector machine (SVM), Bernoulli Naïve Bayes (NB), logistic regression, and linear support vector classification (SVC), specifically tailored for the detection of fake profiles.
Hybrid approach for tweets similarity classification founded on case based reasoning and machine learning techniques Bensassi, Ismail; Kouissi, Mohamed; Ndama, Oussama; En-Naimi, El Mokhtar; Zouhair, Abdelhamid
Bulletin of Electrical Engineering and Informatics Vol 14, No 1: February 2025
Publisher : Institute of Advanced Engineering and Science

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.11591/eei.v14i1.8452

Abstract

Twitter sentiment analysis becomes a popular research subject in the last decade. It aims to extract sentiments of users through their public opinion about a given topic. This article proposes a hybrid approach for Twitter sentiment analysis founded on dynamic case based reasoning (DCBR), multinomial logistic regression machine learning algorithm and multi-agent system. Our approach proposes a method to find similar tweets based on content similarity measure using the scientific measurement of keyword weight term frequency-inverse document frequency (TF-IDF). This approach includes gathering and pre-processing tweets, getting score and polarity of tweets, the use of multinomial logistic regression machine learning algorithm to classify our tweets into various classes, using the feature extraction method to extract useful features and then the K-nearest neighbors (KNN) algorithm to make it easier to find similar tweets to our tweet target case. This approach is adaptive and generic and able to track users' tweet to predict their behavior and sentiments in critical situations and delivering personalized content. The current study focuses on Covid-19 tweets, and a public Twitter dataset is used for this purpose.
Enhancing multi-class text classification in biomedical literature by integrating sequential and contextual learning with BERT and LSTM Ndama, Oussama; Bensassi, Ismail; Ndama, Safae; En-Naimi, El Mokhtar
International Journal of Electrical and Computer Engineering (IJECE) Vol 15, No 4: August 2025
Publisher : Institute of Advanced Engineering and Science

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.11591/ijece.v15i4.pp4202-4212

Abstract

Classification of sentences in biomedical abstracts into predefined categories is essential for enhancing readability and facilitating information retrieval in scientific literature. We propose a novel hybrid model that integrates bidirectional encoder representations from transformers (BERT) for contextual learning, long short-term memory (LSTM) for sequential processing, and sentence order information to classify sentences from biomedical abstracts. Utilizing the PubMed 200k randomized controlled trial (RCT) dataset, our model achieved an overall accuracy of 88.42%, demonstrating strong performance in identifying methods and results sections while maintaining balanced precision, recall, and F1-scores across all categories. This hybrid approach effectively captures both contextual and sequential patterns of biomedical text, offering a robust solution for improving the segmentation of scientific abstracts. The model's design promotes stability and generalization, making it an effective tool for automatic text classification and information retrieval in biomedical research. These results underscore the model's efficacy in handling overlapping categories and its significant contribution to advancing biomedical text analysis.
The impact of BERT-infused deep learning models on sentiment analysis accuracy in financial news Ndama, Oussama; Bensassi, Ismail; Mokhtar En-Naimi, El
Bulletin of Electrical Engineering and Informatics Vol 14, No 2: April 2025
Publisher : Institute of Advanced Engineering and Science

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.11591/eei.v14i2.8469

Abstract

This study delves into the enhancement of sentiment analysis accuracy within the financial news domain through the integration of bidirectional encoder representations from transformers (BERT) with traditional deep learning models, including artificial neural networks (ANN), long short-term memory (LSTM) networks, gated recurrent units (GRU), and convolutional neural networks (CNN). By employing a comprehensive method encompassing data preprocessing, polarity analysis, and the application of advanced neural network architectures, we investigate the impact of incorporating BERT’s contextual embeddings on the models’ sentiment classification performance. The findings reveal significant improvements in model accuracy, precision, recall, and F1 scores when BERT is integrated, surpassing both traditional sentiment analysis models and contemporary natural language processing (NLP) transformers. This research contributes to the body of knowledge in financial sentiment analysis by demonstrating the potential of combining deep learning and NLP technologies to achieve a more nuanced understanding of financial news sentiment. The study’s insights advocate for a shift towards sophisticated, context-aware models, highlighting the pivotal role of transformer-based techniques in advancing the field.
Unified BERT-LSTM framework enhances machine learning in fraud detection, financial sentiment, and biomedical classification Ndama, Oussama; Bensassi, Ismail; Ndama, Safae; En-Naimi, El Mokhtar
IAES International Journal of Artificial Intelligence (IJ-AI) Vol 14, No 6: December 2025
Publisher : Institute of Advanced Engineering and Science

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.11591/ijai.v14.i6.pp5081-5095

Abstract

The current paper proposes a hybrid framework based on the bidirectional encoder representations from transformers (BERT) and long short-term memory (LSTM) networks for classification tasks in three diverse domains: credit card fraud detection (CCFD), financial news sentiment analysis (FNSA), and biomedical paper abstract classification (BPAC). The model leverages the strengths of BERT regarding the learning of contextual embeddings and those of LSTM in capturing sequential dependencies, thus setting the new state-of-the-art performance in each of the three domains. In the CCFD use case, the model was able to achieve an accuracy of 99.11%, considerably outperforming all the competing systems in fraud transaction detection. The BERT-LSTM model achieved a performance of 96.74% for FNSA, improving significantly in sentiment analysis. Finally, the use case of BPAC was robust, with 88.42% accuracy, which clearly classified biomedical abstract sections correctly. It is evident from the findings that this framework generalizes to a wide range of tasks and hence is an adaptable but strong tool in combating challenges of cross-domain classification.