This session consists of a 15-minute talk and a 10 minute Q/A geared toward newcomers to the field, introducing them to the major questions and approaches related to deep learning and inverse problems.

Reliable medical image recovery is crucial for accurate patient diagnoses, but little prior work has centered on quantifying uncertainty when using non-transparent deep learning approaches to reconstruct high-quality images from limited measured data. In this study, we develop methods to address these concerns, utilizing a VAE as a probabilistic recovery algorithm for pediatric knee MR imaging. Through our use of SURE, which examines the end-to-end network Jacobian, we demonstrate a new and rigorous metric for assessing risk in medical image recovery that applies universally across model architectures.

We tackle a challenging blind image denoising problem, in which only single distinct noisy images are available for training a denoiser, and no information about noise is known, except for it being zero-mean, additive, and independent of the clean image. In such a setting, which often occurs in practice, it is not possible to train a denoiser with the standard discriminative training or with the recently developed Noise2Noise (N2N) training; the former requires the underlying clean image for the given noisy image, and the latter requires two independently realized noisy image pair for a clean image. To that end, we propose GAN2GAN (Generated-Artificial-Noise to Generated-Artificial-Noise) method that first learns a generative model that can 1) simulate the noise in the given noisy images and 2) generate a rough, noisy estimates of the clean images, then 3) iteratively trains a denoiser with subsequently synthesized noisy image pairs (as in N2N), obtained from the generative model. In results, we show the denoiser trained with our GAN2GAN achieves an impressive denoising performance on both synthetic and real-world datasets for the blind denoising setting.

Visit the Gather.town to discuss with speakers and other attendees.

We characterize the measurement complexity of compressed sensing of signals drawn from a known prior distribution, even when the support of the prior is the entire space (rather than, say, sparse vectors). We show for Gaussian measurements and \emph{any} prior distribution on the signal, that the conditional resampling estimator achieves near-optimal recovery guarantees. Moreover, this result is robust to model mismatch, as long as the distribution estimate (e.g., from an invertible generative model) is close to the true distribution in Wasserstein distance. We implement the conditional resampling estimator for deep generative priors using Langevin dynamics, and empirically find that it produces accurate estimates with more diversity than MAP.

Visit the Gather.town to discuss with speakers and other attendees.

Visit the gather.town to see the posters.

Visit the Gather.town to discuss with speakers and other attendees.