Biography

My name is Saheed Ganiyu. I am a 4th-year Applied Mathematics PhD student at the University of Arizona with a strong background in Pure and Applied Mathematics, I hold three master's degrees in three different areas of Mathematics (two in Europe and one in the USA—the same place I am currently pursuing my PhD degree). I am doing my research in Machine Learning, where I am working on multi-task learning for better generalization of multiple related tasks while preventing negative transfer. Right from my master degree day, I have developed a keen interest in leveraging analytical and quantitative skills to extract valuable insight from data. I am experienced in data science and machine learning through several university and personal projects.

Academic Status

PhD Student - 4th

Research Area/Department

Applied Mathematics; Data Science; Machine Learning/AI; Mathematics

Major/Specialty

Applied Mathematics

Degrees Earned or in Progress

PhD degree

Academic Preparation

Statistical Machine Learning, SQL/NoSQL Databases, Data Science, Parallel Computing, Stochastic Modeling & Simulation, Numerical Methods for Linear Algebra & Optimization, Big Data Model & Algorithms, Theory of Probability , Linear Algebra, Theory of Statistics, Quantum Mechanics & Computing, Data Analytics & Data Mining, Computational Mathematics, Advanced Regression Analysis, etc.

Research/Publications

Ibrahim, A. A., Ridwan, R. L., Muhammed, M. M., Abdulaziz, R. O., & Saheed, G. A. (2020). Comparison of the CatBoost classifier with other machine learning methods. International Journal of Advanced Computer Science and Applications, 11(11), 738-748. Salaudeen, A., Abubakar, A., & Saheed, G. (2019). Cohort Search, Representation, and Prediction: Application to Medical Data. International Journal of Computer Applications, 975, 8887.

Research/Academic Interests

My research interest lies at the intersection of optimization and machine learning (ML). I have previously worked on optimization methods for solving deep learning (DL) problems. I studied the optimization methods' impacts on deep learning tasks and particularly focused on adaptive learning rate optimizers, where hyperparameter sensitivity, convergence, and generalization were analyzed through an empirical approach. In the course of this project, I gained a strong understanding of gradient-based optimization, convergence and convex analysis, and stochastic dynamics in large-scale learning systems. Moreover, I am currently working on multitask learning (MTL) frameworks. MTL is defined as an approach to which we train a model to carryout many related tasks at a time, instead of training a separate model for each task. My focus is to improve model generalization and prevent negative transfer among related tasks, which is one of the major issues facing MTL. My goal is to come up with principled optimization-based regularization strategies that will allow tasks to share useful representations and keep task-specific distinctions. In addition, I also have a keen interest in machine learning and data science. I have much experience training models for different tasks. I have built a model on high-dimensional spectral data to detect a virus on the surface of a material. This project was challenging and rewarding, as I gained new approaches to solving some unexpected issues from the real data.

Computational and Data Science Areas

Applied Computer Science; Applied Mathematics; Artificial Intelligence and Intelligent Systems; Computer Science; Informatics, Analytics and Information Science; Other Computer and Information Sciences; Statistics and Probability; Visualization and Human-Computer Systems

Motivation

I am highly motivated to be one of the participants in the Sustainable Research Pathways programs, as it is in line with my research interests, academic goals, and ability to apply mathematical and computational methods to impactful, interdisciplinary research. Being a PhD student in Applied Mathematics, and as my research work revolves around optimization and machine learning, where I am currently working on the "theoretical analysis of negative transfer and generalization in a multitask learning framework", having the opportunity to collaborate through SRP-NAIRR, I believe this will bring a valuable opportunity to push these ideas to the real world and a large-scale problem using high-performance computing resources. Moreover, looking beyond the research aspect, I am attracted to the community part of the program—the opportunity to network with scientists, mentors, and peers who have the same passion for scientific innovation and meaningful collaboration. I believe that all these will give me the opportunity to learn from diverse perspectives and be able to contribute my best to work that lifts sustainability and responsible AI research. At the end of this program, I hope to connect with some experienced people who will provide more insights and guidance on my ongoing research (multitasking learning) from their mentorship in interdisciplinary teamwork. I hope to strengthen my research independence and also aim to contribute meaningfully to the SRP-NAIRR community mission.

Lightning Talk Title

Theoretical Analysis of Negative Transfer and Generalization in Multitask Learning.

Keywords (Maximum 20 words)

Multitask Learning; Negative Transfer Learning; Transfer Learning; Optimization; Generalization; Deep Networks; Adaptive Regularization; Artificial Neural Network; Machine Learning; Data Science.