Yes, Generative Models Are the New Sparsity

Given a corrupted or noisy image, how can we improve resolution and obtain faster imaging reconstruction algorithms? Modern deep generative models like GANs have demonstrated excellent performance in representing high-dimensional distributions and therefore “creating” lifelike images. IFML research demonstrates how deep generative models can be used to solve more advanced image reconstruction problems, such as denoising, completing missing data or inpainting, and recovery from linear projections. 

Watch a recent presentation by Alex Dimakis at the Simons Institute where IFML researchers spent a week in residence at the as part of the fall program on Computational Complexity of Statistical Inference (CCSI.) “Yes, Generative Models Are the New Sparsity” presents joint work conducted with Ajil Jalal, Sushrut Kamalkar, Joseph Dean, Glannis Daris, Qi Lei, Ashish Bora, Marius Arvinte, Jon Tamir and Eric Price.

Explore all talks from the Joint IFML/CCSI Symposium.