Doctoral defence: Hassan Abdulgaleel Hassan Salim Eldeeb "Empowering Machine Learning Pipelines with Automated Feature Engineering“

On 8 October at 10:15, Hassan Abdulgaleel Hassan Salim Eldeeb will defend his thesis "Empowering Machine Learning Pipelines with Automated Feature Engineering“ to obtain the degree of Doctor of Philosophy (in Computer Science).

Supervisor:
Associate Professor Radwa Mohamed El Emam El Shawi, University of Tartu

Opponents:
Professor Ladjel Bellatreche, National Engineering School for Mechanics and Aerotechnics (France) 
Professor Çağatay Çatal, Qatar University (Qatar)

Summary
In the era of Big Data, Automated Machine Learning (AutoML) has emerged as a transformative force, bridging the gap between burgeoning data volumes and the limited supply of data science expertise. This technological leap is epitomized by frameworks that automate the complex and iterative process of crafting machine learning models, thereby broadening the accessibility and efficiency of data analysis. 

Central to this evolution is the innovation in Automated Feature Engineering (FE) — a pivotal yet traditionally labor-intensive step that enhances model accuracy through the intelligent selection and transformation of data attributes. Herein lies the significance of BigFeat, a pioneering framework meticulously engineered to automate and revolutionize the FE process. BigFeat distinguishes itself by seamlessly blending scalability with interpretability, ensuring the automated generation of features does not obscure the rationale behind model predictions. 

Through exhaustive benchmarking against existing frameworks, BigFeat has consistently demonstrated superior performance, delivering significant accuracy improvements across diverse datasets. Its design, focused on dynamic feature generation and selection, not only boosts model performance but also preserves the interpretability essential for user trust and understanding. 

BigFeat's introduction marks a significant milestone in the AutoML landscape, offering a robust solution that democratizes advanced data analysis. By mitigating the need for extensive domain knowledge in FE, BigFeat empowers a wider spectrum of users to leverage machine learning, fostering innovation and insight across various fields.