Uncertainty Quantification with Generative Models
arXiv:1910.10046
Abstract
We develop a generative model-based approach to Bayesian inverse problems, such as image reconstruction from noisy and incomplete images. Our framework addresses two common challenges of Bayesian reconstructions: 1) It makes use of complex, data-driven priors that comprise all available information about the uncorrupted data distribution. 2) It enables computationally tractable uncertainty quantification in the form of posterior analysis in latent and data space. The method is very efficient in that the generative model only has to be trained once on an uncorrupted data set, after that, the procedure can be used for arbitrary corruption types.
accepted submission to the Bayesian Deep Learning NeurIPS 2019 Workshop
References in corpus (11)
- Auto-Encoding Variational Bayes
- Deep Convolutional Neural Network for Inverse Problems in Imaging
- Deep Image Prior
- Wasserstein Auto-Encoders
- Image Super-Resolution Using Deep Convolutional Networks
- Recurrent Inference Machines for Solving Inverse Problems
- Deep Bayesian Inversion
- Posterior inference unchained with EL_2O
- Leveraging the Exact Likelihood of Deep Latent Variable Models
- Unsupervised Data Imputation via Variational Inference of Deep Subspaces
- Conditional Inference in Pre-trained Variational Autoencoders via Cross-coding
Cited by in corpus (11)
- A Review of Uncertainty Quantification in Deep Learning: Techniques, Applications and Challenges
- Towards Precise and Accurate Calculations of Neutrinoless Double-Beta Decay: Project Scoping Workshop Report
- PID-GAN: A GAN Framework based on a Physics-informed Discriminator for Uncertainty Quantification with Physics
- Probabilistic Autoencoder
- Nonlinear 3D Cosmic Web Simulation with Heavy-Tailed Generative Adversarial Networks
- A Generative Model of Galactic Dust Emission Using Variational Inference
- Bridging the Gap Between Explainable AI and Uncertainty Quantification to Enhance Trustability
- A Probabilistic Autoencoder for Type Ia Supernovae Spectral Time Series
- Dense Uncertainty Estimation
- Fully Unsupervised Diversity Denoising with Convolutional Variational Autoencoders
- Bayesian Reasoning with Trained Neural Networks