Mahalakshmi Sabanayagam

Mahalakshmi Sabanayagam is a Researcher at the Technical University of Munich (TUM), affiliated with the School of Computation, Information and Technology and the Department of Theoretical Foundations of Artificial Intelligence. Her research focuses on the theory of deep learning and graph-based learning, with notable contributions to adversarial robustness, neural network certification, and graph representation learning. She currently serves as a Teaching Assistant for the course 'Gems of Informatics 3 (IN2176): Modelling and analysis of real-world graphs.'

Her work bridges theoretical foundations and practical applications, including analysis of neural tangent kernels in graph networks, robustness certification against adversarial attacks, and cross-domain machine learning techniques. Key research directions include understanding decision boundaries via Hessian analysis, tensorized autoencoder architectures, and graphon-based network clustering.

Publications span topics from adversarial machine learning to interdisciplinary applications in physics and finance. While currently no awards or grants are listed, her active publication record reflects ongoing contributions to artificial intelligence theory and graph-based methodologies. She is part of Prof. Debarghya Ghoshdastidar's research group at TUM's Faculty of Computer Science.

Lab affiliations include the Theoretical Foundations of Artificial Intelligence (I7) group, focusing on rigorous mathematical analysis of AI systems and their real-world graph applications. No advising relationships or student collaborations are explicitly documented in available materials.