Phd Student at Ferhat Abbas University, Algeria
Mafaza Chabane is a passionate researcher and developer currently pursuing her Ph.D. at the University of Ferhat Abbas Sétif 1, Algeria, focusing on Robust Natural Language Understanding. 📚 With a deep commitment to AI and computational linguistics, she has authored a Q1 journal article and contributed to several international conferences. 💡 Her master’s dissertation explored transfer learning and transformers for multilingual text classification, securing an excellent grade. She has consistently ranked among the top students and excelled in both her undergraduate and master’s programs. Beyond academia, Mafaza has also worked as a freelance mobile app developer and held a position at ITSourceOne in Jordan. 💼 Her skills span across deep learning, natural language processing, mobile app development using Flutter, and UI/UX design. 🚀 Proficient in Arabic, French, and English, she stands out as a multilingual communicator and problem-solver. 🔧 Mafaza blends academic depth with technical versatility to address complex real-world challenges.
Professional Profile
Orcid
Scopus
Google Schloar
🎓 Education
Mafaza’s academic path reflects unwavering excellence and intellectual curiosity. 🌟 She began her academic journey at Fatima El-Zahraa High School with a solid score of 15.03/20, which laid the foundation for a remarkable higher education track. During her Bachelor’s (2016–2019) at University of Ferhat Abbas Sétif 1, she ranked 3rd out of 222, showcasing early signs of academic brilliance. 📈 Her Master’s (2019–2021) focused on machine learning, deep learning, and NLP, culminating in a dissertation on text classification using transformers that earned an excellent grade (18/20). She maintained the top class rank for four semesters, achieving an overall grade of 15.69/20. 📊 Presently, as a Ph.D. candidate, she is delving into low-resource language understanding, tackling vital challenges in Arabic dialect processing. Her educational path reflects not only top performance but also a clear focus on innovative, real-world AI problems—positioning her as a strong, research-driven academic. 📘🧠
💼 Professional Experience
Mafaza has seamlessly combined academic knowledge with practical industry skills. From 2021 to 2023, she worked remotely with ITSourceOne in Jordan, contributing to innovative tech projects. 🇯🇴 In parallel, since 2020, she has been an active freelancer in Setif, developing mobile applications across a wide range of domains, including education, food delivery, language translation, and healthcare. 📱💡 Notably, she launched and contributed to platforms like Ennajah QCM, Floppy Delivery, Yumi Food, and Private Call Translation, many of which are published on Google Play and the App Store. Her stack includes Flutter, Firebase, and both SQL and NoSQL databases, demonstrating end-to-end mobile development proficiency. 👩💻 Her problem-solving mindset and grasp of design architectures (MVC-MVVM) underscore her capability in building scalable, user-centric applications. Through this, Mafaza has established herself as a well-rounded tech expert who integrates software engineering with AI research. 🔧📲
🔬 Research Interests
Mafaza’s core research lies at the intersection of natural language processing, low-resource languages, and deep learning. 🧠 Her academic work began with transfer learning and deep neural networks to improve text classification and translation, particularly for Arabic dialects. She has since expanded her focus to dialect identification, sentiment analysis, and speech emotion recognition, aiming to improve computational understanding of underrepresented languages. 🌍 She integrates transformers (e.g., CAMeLBERT), FastText, and hybrid neural models to boost performance across tasks such as disaster event classification, medical image recognition, and language identification. Her publication in Expert Systems with Applications (Q1) stands as a testament to her innovative contributions in low-resource NLP. 📄🧾 Additionally, her involvement in image-based AI tasks (e.g., cancer detection and brain tumor classification using CT/MRI scans) reflects multidisciplinary interest, blending language technologies with healthcare AI and disaster management systems. 🔬🌐
🏆 Awards and Honors
Mafaza’s academic path is decorated with exceptional honors that validate her potential. 🌟 She was ranked top of her class for four semesters during her master’s studies—an outstanding recognition of sustained academic brilliance. 📚 She also secured 3rd rank out of 222 students in her undergraduate program, which speaks volumes of her consistency and competitiveness. Her master’s dissertation was awarded an excellent distinction (18/20), reviewed by an academic jury. 🥇 While formal research awards may still be in the early stage of her career, her Q1 journal publication and regular conference presentations at top-tier platforms such as ICTDM, ISNIB, and PAIS are indicators of growing international recognition. 🎤 Her work is increasingly being cited and appreciated in fields of NLP and deep learning, and she is actively engaged in collaborative interdisciplinary research, making her a promising candidate for future academic and industry accolades. 🏅📝
📚 Publications Top Note
1. COVID-19 Detection from X-ray and CT Scans Using Transfer Learning
-
Authors: M. Berrimi, S. Hamdi, R.Y. Cherif, A. Moussaoui, M. Oussalah, M. Chabane
-
Year: 2021
-
Citations: 44
-
Source: 2021 International Conference of Women in Data Science at Taif University (WiDS)
-
Summary:
This study applies transfer learning to detect COVID-19 using chest X-ray and CT scans. Several pre-trained deep learning models were fine-tuned and evaluated for performance in binary and multi-class classification. The results indicated that transfer learning significantly improves detection accuracy with minimal data and training resources.
2. Ensemble Transfer Learning for Improved Brain Tumor Classification in MRI Images
-
Authors: S. Hamdi, A. Moussaoui, M. Berrimi, A. Laouarem, M. Chabane
-
Year: Not officially published (early access or conference proceedings)
-
Citations: 1
-
Source: Unspecified (likely conference/workshop or preprint)
-
Summary:
This work explores an ensemble-based transfer learning approach using multiple deep CNN architectures for the classification of brain tumors from MRI images. It focuses on boosting accuracy and robustness through combining different model predictions.
3. Advancing Low-Resource Dialect Identification: A Hybrid Cross-Lingual Model Leveraging CAMeLBERT and FastText for Algerian Arabic
-
Authors: M. Chabane, F. Harrag, K. Shaalan
-
Year: 2025
-
Citations: Not yet cited (new)
-
Source: Expert Systems with Applications, Volume 284, Article 127816
-
Summary:
This article introduces a hybrid cross-lingual model for dialect identification, focusing on Algerian Arabic. It combines CAMeLBERT (a contextualized BERT model for Arabic) with FastText embeddings to address low-resource linguistic settings, achieving high performance in dialect recognition.
4. Bridging the Gap: Transfer Learning for Dialect Identification in Low-Resource Settings – A Case Study with Algerian Arabic
-
Authors: M. Chabane, F. Harrag, K. Shaalan, S. Hamdi
-
Year: 2025
-
Citations: Not yet cited (recent)
-
Source: 2025 International Symposium on iNnovative Informatics of Biskra (ISNIB)
-
Summary:
The study investigates transfer learning techniques tailored for dialect identification, focusing on Algerian Arabic as a case study. It evaluates several pre-trained multilingual models and proposes strategies for improving performance on underrepresented dialects.
5. SECA-Net: A Lightweight Spatial and Efficient Channel Attention for Enhanced Natural Disaster Recognition
-
Authors: S. Hamdi, A. Moussaoui, M. Chabane, A. Laouarem, M. Berrimi, M. Oussalah
-
Year: 2024
-
Citations: Not yet cited
-
Source: 2024 International Conference on Information and Communication Technologies (ICICT)
-
Summary:
Proposes SECA-Net, a deep neural network architecture integrating spatial and channel attention mechanisms to detect natural disasters from satellite images. The model emphasizes low computational complexity with improved feature representation.
6. Beyond Deep Learning: A Two-Stage Approach to Classifying Disaster Events and Needs
-
Authors: M. Chabane, F. Harrag, K. Shaalan
-
Year: 2024
-
Citations: Not yet cited
-
Source: 2024 International Conference on Information and Communication Technologies (ICICT)
-
Summary:
This paper introduces a two-stage pipeline for natural disaster classification: the first stage identifies the disaster event type, while the second detects the specific needs or requests. It uses traditional ML and deep learning for complementary performance.
7. Uncovering Linguistic Patterns: Exploring Ensemble Learning and Low-Level Features for Identifying Spoken Arabic, English, Spanish, and German
-
Authors: S. Hamdi, A. Moussaoui, M. Chabane, A. Laouarem, M. Berrimi, M. Oussalah
-
Year: 2023
-
Citations: Not yet cited
-
Source: 2023 5th International Conference on Pattern Analysis and Intelligent Systems (PAIS)
-
Summary:
Investigates spoken language identification using an ensemble learning framework that fuses low-level audio features (MFCC, spectral) for recognizing multiple languages. The system targets Arabic, English, Spanish, and German, achieving high multi-lingual classification accuracy.
🔚 Conclusion
Mafaza Chabane is an emerging force in AI research, blending deep theoretical knowledge with real-world impact. 💼 From pioneering dialect recognition techniques to deploying mobile applications that meet local community needs, she represents the future of ethical and inclusive AI. 🔍 Her trilingual communication skills, cross-domain expertise, and research in natural language understanding for low-resource settings position her uniquely in today’s global AI landscape. 🌐 With a foot in both academia and industry, she brings versatility, innovation, and a forward-thinking mindset. Her consistent excellence across education, hands-on development, and research makes her a strong contender for any “Best Researcher” or “Young Innovator” award. 🏆 As she continues her Ph.D. and broadens her contributions, Mafaza is set to make even more significant marks in computational linguistics, AI for social good, and language technologies for the underserved. 🚀🌟