Abdel Ghani Labassi

Bio

I am currently pursuing my PhD in the Applied Mathematics and Statistics Department at The Johns Hopkins University, Baltimore, Maryland, USA. My research focuses on optimization and machine learning, particularly in developing new methods to solve large-scale problems efficiently, under the supervision of Mateo Díaz.

I completed my MSc in Computer Science in the Department of Copmuter Science and Operations Research at Université de Montréal, Québec, Canada, where I worked with Andrea Lodi and Didier Chetelat. In addition, I obtained a double major BSc in Computer Science and Mathematics from the same department. During my undergraduate studies, I gained industry experience as a Software Engineer Intern at Noviflow Inc. in Montreal, working on web app development and data visualization projects. I also contributed to machine learning engineering tasks as an intern at the CERIST Agency in Algeria.

Research

My research focuses on optimization and statistics for machine learning. Some of my current projects include:

• Inverse Optimization Techniques for Large Language Model Interpretability: Investigating new techniques to improve the interpretability of large models.
• Novel Optimization Algorithm for Composite Machine Learning: Developing new algorithms to improve the scalability of machine learning problems.

Publications

• Learn to Compare Nodes in Branch and Bound with GNN, A.G.Labassi, Didier Chetelat, Andrea Lodi. Presented at NeurIPS, 2022.

Teaching

I have been a teaching assistant for the following courses:

• Optimization for Finance - Johns Hopkins University, 2022-2023
• Network Flow Optimization - Johns Hopkins University, 2022-2023
• Intro to Algorithmics - Université de Montréal, 2019-2022

Projects

• Kaggle Reddit Comment Classification Competition - Led a team that ranked 4th out of 103 teams using a BERT model.
• Unsupervised Tweet Clustering for Mental Health - Collaborated with Health Canada to detect signs of depression and suicidal tendencies from tweets.

Contact

If you would like to get in touch, feel free to reach out via email at alabass1@jhu.edu or connect with me on LinkedIn.