# Including Physics in Deep Learning -- An example from 4D seismic pressure saturation inversion

## TL;DR

This paper demonstrates how incorporating physics-based priors and noise injection techniques into neural networks improves seismic pressure saturation inversion, effectively handling data uncertainty and anomalies.

## Contribution

It introduces a method to embed physical information into neural network architecture and employs noise injection during training for better real-world data transfer.

## Key findings

- Physics-informed neural networks enhance seismic inversion accuracy.
- Noise injection improves model robustness to field data.
- Method effectively handles imbalanced and sparse data scenarios.

## Abstract

Geoscience data often have to rely on strong priors in the face of uncertainty. Additionally, we often try to detect or model anomalous sparse data that can appear as an outlier in machine learning models. These are classic examples of imbalanced learning. Approaching these problems can benefit from including prior information from physics models or transforming data to a beneficial domain. We show an example of including physical information in the architecture of a neural network as prior information. We go on to present noise injection at training time to successfully transfer the network from synthetic data to field data.