NEWS
4th batch of Master’s in Big Data Science & Analytics Gradates Studies at UOB Explore Medical and Environmental Solutions Using Big Data
Sakhir – Bahrain University (Khadijah Abdusalam)
15 July 2024
The fourth batch of students recently graduated from the Master’s program in Big Data Science and Analytics, which is offered in cooperation between the College of Science and the College of Information Technology at the University of Bahrain, bringing the number of graduates in the program to 40 students so far.
Three theses were discussed during June 2024, focusing on practical applications in the field of medicine and diagnosis, to natural disaster predictions.
Student Ziad Atef Allam presented a study titled “Enhancing Multiple Bends in Collecting Magnetic Resonance Images Using Aggressive Learning”. The study was supervised by Dr. Riadh BinMohammed Ksantini, Associate Professor at the College of Information Technology, and Dr. Sawsan Hilal, Head of the Mathematics Department and Coordinator of the Master’s Program in Big Data Science and Analytics.
Allam proposed a new model for AdversFAE based on the use of discordantly restricted interpolation in the assembly stage, to escape the development of features in automatic encoders. The study applied the proposed model to six datasets of the MedMNIST v2 set of medical datasets ranging from X-rays to CT scans, and the proposed model proved effective as it outperformed the ConvDynAE in terms of accuracy and (F1_score) in assembly.
In addition, student Hamad Munir Malik presented a study titled ” A Novel Model for Chart-to-Text Generation by Utilizing Neural Network Models”. The study was supervised by Prof. Nabil Mahmood Hewahi from the College of Information Technology.
The study developed a model to promote automatic text creation from scientific charts, where the neural network (CNN) was used to extract features from the chart, and the neural network (RNN) to create accurate descriptive captions. The research found that Bi-LSTM, a type of RNN, is the best, allowing the model to capture the past and future context of the input sequence.
The student, Fathan Tael Shajara, presented a study titled “Earthquake Prediction Using Machine Learning Approach”, supervised by Prof. Nabil Mahmood Hewahi from the College of Information Technology.
The study looked at predicting earthquake characteristics using machine learning techniques, and used USGS data to predict earthquake magnitude, and Los Alamos National Laboratory (LANL) data to predict when earthquakes would occur. The results recommended focusing on improving feature engineering and selection processes, and integrating various other additional indicators such as seismic activity, historical patterns, and geological data, to further improve the prediction capabilities of these models.
