NLP Inspired Training Mechanics For Modeling Transient Dynamics

TL;DR

This paper applies NLP-inspired training techniques like teacher forcing and curriculum learning to improve the accuracy and robustness of ML models simulating transient dynamics, especially in vortical flows.

Contribution

It introduces NLP-inspired training mechanics to enhance ML models for transient dynamics, achieving over 50% accuracy improvements.

Findings

Enhanced accuracy of ML models by more than 50%.

Improved robustness and generalizability in simulating vortical flows.

Demonstrated effectiveness of NLP-inspired training in physics-based modeling.

Abstract

In recent years, Machine learning (ML) techniques developed for Natural Language Processing (NLP) have permeated into developing better computer vision algorithms. In this work, we use such NLP-inspired techniques to improve the accuracy, robustness and generalizability of ML models for simulating transient dynamics. We introduce teacher forcing and curriculum learning based training mechanics to model vortical flows and show an enhancement in accuracy for ML models, such as FNO and UNet by more than 50%.