Prasad Krishnan is with the Signal Processing and Communications Research Centre, International Institute of Information Technology Hyderabad (IIIT Hyderabad) as a faculty member. He received the B.E. degree from the College of Engineering at Guindy, Anna University, in 2007, and the Ph.D. degree from the Department of Electrical Communication Engineering, Indian Institute of Science, Bangalore, in 2014. His research has been supported through generous grants from the Govt. of India (SERB/ANRF), as well as private agencies like Qualcomm. His current research interests include but is not limited to Algebraic Coding Theory, Private Information Retrieval, and Information-theoretic limits of DNA data storage. 

Abstract: In Private Information Retrieval, a client seeks to download a desired file privately from a library stored in a distributed fashion across multiple databases (or servers). Characterisation of the capacity (i.e., the largest possible rate) of PIR under perfect information-theoretic privacy, for various typical distributed storage settings, has been the primary focus of the last decade in IT-PIR. This session will present results regarding some recent shades of IT-PIR, including PIR under relaxed notions of privacy, PIR under special storage schemes, with constraints on required file-sub-packetization, etc. These explorations lead to newer questions that are often simple to state, but perhaps not so simple to solve.

Parthe Pandit is the Thakur Family Chair Assistant Professor with the Center for Machine Intelligence and Data Science (C-MInDS) at IIT Bombay. He was a Simons Postdoctoral Fellow at UCSD, and obtained his PhD in ECE, MS in Statistics both from UCLA, and his undergraduate degrees from IIT Bombay. He is a recipient of the Schmidt Sciences AI2050 Early Career Fellowship (2024), and the Jack K Wolf ISIT student paper award (2019).

Abstract: Modern machine learning involves design choices about the architecture of the model, optimization procedures used for training, and amount of data. These choices interact in a complex manner. However, to build *good* ML systems, it is essential to understand their effect on the performance of the system. Through this tutorial, we will build a framework to analyze some of these design choices. We will focus on addressing issues arising out of the models being overparameterized and overfitted, which have challenged classical statistical learning theory based on uniform convergence and have demanded the development of a revised theory of statistical ML. While this is a fast growing area with many new tools introduced to pose and answer important questions, we will focus on some techniques based on spectral analysis of kernel methods, and nonlinear statistical inverse problems in high dimensions. We will conclude by outlining some important open questions in the field.

Emina Soljanin is a Distinguished Professor of Electrical and Computer Engineering at Rutgers. Before moving to Rutgers in January 2016, she was a (Distinguished) Member of Technical Staff for 21 years in the Mathematical Sciences Research Center of Bell Labs. Her interests and expertise are broad, currently spanning distributed computing and quantum information science. She is an IEEE Fellow, an outstanding alumnus of the Texas A&M School of Engineering, the 2011 Padovani Lecturer, a 2016/17 Distinguished Lecturer, and the 2019 IEEE Information Theory Society President. In 2023, Emina received the IEEE Information Theory Society Aaron D. Wyner Distinguished Service Award and Mrs. Urmila Agrawal Distinguished Visiting Chair Professorship at the Indian Institute of Science.

Abstract: Quantum information science is a rapidly advancing and inherently interdisciplinary field, making it both compelling and challenging for newcomers. This tutorial provides an accessible introduction to the essentials of quantum information processing by first addressing three foundational questions: How is quantum information represented? How is it processed? How is classical information extracted from quantum states? We then present several fundamental quantum algorithms and protocols that illustrate the distinctive power of quantum computing and conclude with examples that highlight the role of quantum correlations. The tutorial emphasizes reliability and security topics, including quantum error correction, quantum key distribution, the quantum one-time pad, quantum money, and quantum games. The only prerequisites are undergraduate probability and algebra.

Saket Anand is an Associate Professor at IIIT-Delhi. He completed his PhD from Rutgers University in 2013. His research interests span geometric computer vision, semi-supervised and unsupervised learning, and robust statistical methods for computer vision. He has been applying computer vision techniques  to domains like wildlife conservation and conflict management, road safety and autonomous driving, and remote sensing for agriculture. In addition to publishing and reviewing papers at venues like ECCV, CVPR, TPAMI, etc., he has served as an Area Chair for IEEE CVPR 2026, ICCV 2025, WACV 2020, 2023, 2024 and ICVGIP 2020/21, the premier Indian Vision Conference, and as a Program co-Chair for IEEE WACV 2022.

Lokender Tiwari is the Head of Computer Vision and AI at BotLab Dynamics, where he leads the development of next-generation intelligent drones for defense applications. He completed his PhD from IIIT-Delhi. His research interests include 3D vision and spatial AI, autonomous driving & flying, sensor fusion and signal intelligence, neural physics-based simulation, and differentiable graphics for physics-aware immersive technologies.  He has been applying above techniques to build multi-modal systems for comprehensive situational awareness, autonomous navigation, real-time detection & tracking, generative simulations etc. His work has been published at top-tier conferences such as ECCV, CVPR, ICCV, ISMAR etc. He also serves as a reviewer and program committee member for conferences and journals like CVPR, ICLR, ICML, ECCV, AAAI, TPAMI, IJCV etc.

Abstract: Following the remarkable success of deep learning on visual recognition and related 2D vision tasks, there has been an increasing interest in developing deep learning models for 3D vision problems. In this tutorial, we will focus on multi-view geometry problems and study some of the traditional and modern, learning-based solutions. We will begin with a brief review of the fundamentals of image formation and camera geometry before formulating the related problems of 3D reconstruction, Structure from Motion (SfM) and Simultaneous Localization and Mapping (SLAM), followed by an overview of traditional computer vision solutions and their relevance today. We will then review some of the key ideas driving learning-based solutions and introduce some recently proposed models for 3D vision. We will conclude with practical considerations for tackling 3D vision problems today and some open research questions. The prerequisites for this tutorial are undergraduate linear algebra and a working knowledge of  PyTorch and deep learning models (CNNs and Transformers).