Skip to content

01

Last Updated: 10 months ago2024-05-12 ; Contributors: AhmedThahir, web-flow

Typical Scientific Tasks

typical_scientific_tasks

PINN

Physics-Informed Neural Networks

image-20240227001251022

Digital Twin

Digital_Twin

image-20240310000827496

Physics’ Properties Preferred in Machine Learning

  • Interpretability

  • Generalizability

  • Parsimony/Simplicity

  • Symmetries/Conservations

  • Invariance: (fg)(x)=f(x)(f \circ g)(x) = f(x)

  • Equivariance: (fg)(x)=(gf)(x)(f \circ g)(x) = (g \circ f)(x)

where

  • ff is deep learning model
  • gg is a transformation such as rotation

Comments