Home

Biography

Abdul Quadir is a Ph.D. candidate in the Department of Mathematics at the Indian Institute of Technology (IIT) Indore, advised by Prof. M. Tanveer. His research lies at the intersection of statistical learning theory and deep learning, with a focus on developing robust, theoretically grounded algorithms for complex and uncertain data environments. He earned his M.Sc. in Mathematics from IIT Indore in 2022 and his B.Sc. (Hons.) in Mathematics from Doon University in 2020, graduating with the University Gold Medal as the top-ranking student in his cohort.

His foundational research investigates the theoretical and computational foundations of hyperplane-based discriminative learning, with emphasis on formulating mathematically rigorous optimization problems deeply rooted in statistical learning theory. By carefully balancing the principles of empirical risk minimization, regularization theory, and convex optimization, his work establishes a unified mathematical foundation that is both theoretically sound and computationally efficient for real-world supervised learning tasks.

Leveraging the granular computing framework, he designed scalable and robust classifiers built upon region-level geometric abstractions, delivering improved noise tolerance, computational efficiency, and generalization performance. His contributions to randomized neural network architectures focus on random vector functional link networks and broad learning systems, where he introduced principled modifications grounded in graph embedding, multi-view fusion, and novel loss functions — addressing fundamental limitations of shallow learning and significantly enhancing robustness, expressiveness, and generalization capability.

Taken together, these contributions brought into sharp focus a deep and fundamental limitation of existing learning frameworks — their inability to adequately capture the rich structural dependencies, hierarchical feature interactions, and complex relational geometry that characterize real-world high-dimensional and graph-structured data. This critical observation serves as a strong theoretical motivation, driving the development of mathematically principled and structurally expressive learning frameworks.