Computational Imaging (CI) combines advanced computer algorithms with imaging sensors to capture and interpret the physical world. This interdisciplinary field has diverse applications, including smartphone cameras, AR/VR displays, autonomous driving, computed tomography (CT), magnetic resonance imaging (MRI), microscopy, and even black hole imaging. In this course, we present a unified framework for a wide range of CI problems through the lens of inverse problems. Students will explore both classic and modern algorithms for solving these problems and apply them to real-world imaging challenges. The course is structured as follows. Part I:  Inverse Problems, Bayesian Inference, Image Priors. Part II: Maximum A Posteriori (MAP) Estimation, Optimization-based Methods, Gradient Descent, Proximal Gradient Method, and Alternating Directions Method of Multipliers. Part III: Minimum Mean Squared Error (MMSE) Estimation, Monte Carlo Method, Variational Autoencoder (VAE), Score-Matching Langevin Dynamics, Diffusion Models. The course will also cover the basics of optimization and image processing, providing essential foundational knowledge.

Prerequisites: Linear Algebra, Probability, Signal and Systems & Machine Learning

Preferred Background knowledge: Real Analysis, Optimization & Measure Theory (a touch).