In this talk, I will present three research vignettes spanning explainability, model editing, and quantum generative modeling. First, I will discuss a compelling problem in explainability that emerged from our experience deploying XraySetu—a diagnostic service launched during the COVID-19 pandemic. Second, I will describe our work on identifying structural components for model editing. Notably, our approach does not require access to the original training data or knowledge of the loss functions used to train the model. Instead, it hinges on a novel lower bound on the total variation (TV) distance via witness functions—a result that is of independent interest and, for example, enables upper bounds on the Bayes error rate for the Fisher Discriminant. Finally, time permitting, I will present ongoing work on learning algorithms for generative AI models in a quantum computing setting. In particular, I will outline our derivation of Minorant Maximization–style algorithms, in the spirit of Expectation-Maximization (EM), for large-scale transverse-field Ising models.

Short Bio:

Chiranjib Bhattacharyya is currently Professor in the Department of Computer Science and Automation, Indian Institute of Science. His research interests are in foundations of Machine Learning, Optimisation and their applications to Industrial problems. He is fellow of Indian Academy of Engineering, Indian Academy of Sciences and Asia-Pacific Artificial Intelligence Association. He joined the Department of CSA, IISc, in 2002 as an Assistant Professor. Prior to joining the Department he was a postdoctoral fellow at UC Berkeley. He holds BE and ME degrees, both in Electrical Engineering, from Jadavpur University and the Indian Institute of Science, respectively, and completed his PhD from the Department of Computer Science and Automation, Indian Institute of Science.