Bayesian Physics Informed Neural Networks for Linear Inverse problems

TL;DR

Inverse problems are often ill-posed and require uncertainty-aware solutions. The authors propose a Bayesian PINN (BPINN) framework that integrates Bayesian inference with physics-informed neural networks to recover unknowns from indirect measurements, with MAP appearing as a deterministic limit. In the linear Gaussian setting, the posterior over the unknowns is Gaussian, and the NN is trained via supervised or unsupervised data using losses that fuse data fidelity, forward-model consistency, and prior information. The work discusses practical challenges such as prior choice, neural network architecture, optimization stability, and computational cost, and points toward real-time inversion with uncertainty quantification in imaging and related fields.

Abstract

Inverse problems arise almost everywhere in science and engineering where we need to infer on a quantity from indirect observation. The cases of medical, biomedical, and industrial imaging systems are the typical examples. A very high overview of classification of the inverse problems method can be: i) Analytical, ii) Regularization, and iii) Bayesian inference methods. Even if there are straight links between them, we can say that the Bayesian inference based methods are the most powerful, as they give the possibility of accounting for prior knowledge and can account for errors and uncertainties in general. One of the main limitations stay in computational costs in particular for high dimensional imaging systems. Neural Networks (NN), and in particular Deep NNs (DNN), have been considered as a way to push farther this limit. Physics Informed Neural Networks (PINN) concept integrates physical laws with deep learning techniques to enhance the speed, accuracy and efficiency of the above mentioned problems. In this work, a new Bayesian framework for the concept of PINN (BPINN) is presented and discussed which includes the deterministic one if we use the Maximum A Posteriori (MAP) estimation framework. We consider two cases of supervised and unsupervised for training step, obtain the expressions of the posterior probability of the unknown variables, and deduce the posterior laws of the NN's parameters. We also discuss about the challenges of implementation of these methods in real applications.