Welcome to the Digital Repository of Mohamed Boudiaf University - M'Sila

The University of Mohamed Boudiaf - Institutional Repository (UMB-IR) is your one-stop shop for the university's research findings. This digital archive preserves a wide range of scholarly materials, including publications, conference presentations, book contributions, entire books, theses, and research data. Data can be in various formats, such as videos, spreadsheets, code, and images. The IR also keeps presentations and other scholarly resources.

 

Communities in University of M'Sila Repository

Select a community to browse its collections.

Now showing 1 - 10 of 11

Recent Submissions

ItemOpen Access
On the Vibrations of Traveling Strings and Membranes
(Mohamed Boudiaf University of M'sila, 2024-06-10) Nour, Bourenane; Supervisor: Abdelmouhcene, Sengouga
In this study, we examine the one-dimensional and two-dimensional wave equations in time-dependent domains. We derive the wave equation using Hamilton’s principle. Then, we employ the method of separation of variables to solve the equation. We also consider the case with interior linear damping.
ItemOpen Access
Coverage Optimization in Wireless Sensor Networks Using Memetic Algorithm
(Mohamed Boudiaf University of M'sila, 2025-06-15) Chaima, Bougoutaia; Takoua, Khelifi; Supervisor: Raouf Ouanis, Lakehal Ayat
This dissertation addresses the coverage optimization problem in Wireless Sensor Networks (WSNs), a critical factor affecting network efficiency and sustainability. A hybrid memetic algorithm is proposed, combining global and local search strategies to optimize sensor de ployment and reduce overlap while maintaining maximum coverage with minimal sensor us age. A precise mathematical model is formulated, and simulation experiments are conducted to evaluate the performance of the proposed approach. Results show significant improve ments in coverage quality and energy efficiency compared to conventional algorithms
ItemOpen Access
Skin Diseases Detection From Images
(Mohamed Boudiaf University of M'sila, 2025-06-15) Amira, Lamrour; khadidja, Denidni; Supervisor: Roussafi, Mahdjoubi
Many skin diseases are a big problem for public health since it can be hard to identify them early on. Some conventional diagnosis methods such as inspecting a patient, are not always perfect and may take a long time. Recent years have see impressive developments in artificial intelligence, especially in deep learning, helping to bring about new possibilities in medicine. In our study, a deep learning system is suggested to assist in spotting skin diseases early by studying skin images.
ItemOpen Access
Automated Program Repair using Large Language Models (LLMs)
(Mohamed Boudiaf University of M'sila, 2025-06-15) Amina, MEKIDECHE; Amani, DJAIDJA; Supervisor: Hichem, DEBBI
This work proposes an approach to APR by integrating a fine-tuned CodeLLaMA-7b model with a GraphRAG-based retrieval framework. We first applied LoRA to fine-tune the CodeLLaMA model using the domain-specific RepairLLaMA dataset. To enhance contextual awareness, we built a graph-based retriever that combines CodeBERT-generated embeddings with structural relationships between buggy and fixed code pairs from the Defects4J dataset. Our system prioritizes relevant repair examples and enables efficient retrieval of code context. Additionally, we developed a simple web interface to provide real-time bug fix suggestions, demonstrating the practical applicability of our pipeline.
ItemOpen Access
Forecasting Volatility of Cryptocurrencies Using Machine Learning and Deep Learning
(Mohamed Boudiaf University of M'sila, 2025-06-15) Mohammed Elamin, Amari; Supervisor: Lamri, Sayad
This thesis investigates the application of machine learning and deep learning methods into forecasting cryptocurrencies volatility. Using Bitcoin and Ethereum datasets, the performance of the following models was analyzed: Random Forest, XGBoost, LSTM, and GRU. The results show that deep learning models, specifically LSTM, outperforms machine learning models on every metric. These findings suggest that RNN models are better equipped to handle temporal dependencies in time series data that have sequential nature.