Eco SystemThe relationship between the natural sciences and Data Science has a long history.
Machine Learning has also entered its history . Research is progressing to apply machine learning methods to conventional research in the natural sciences.
Neural Network (Deep Learning) have the characteristic of being able to approximate the mathematical formulas that represent any complex data.
In the natural sciences, we discover and use simple and easy-to-understand mathematical formulas, but if it is more important to use them, neural networks are a good way to go.
Physics-Informed Neural Network is a method to create a highly accurate model of a neural network by including the formula of the laws of physics in the formula of the neural network.
For simulations that deal directly with the equations of the laws of physics, this approach may be better when the computation is huge.
There is also an approach that utilizes machine learning for efficient use of huge amounts of data.
In materials informatics, we find mathematical formulas that can express the whole of a lot of experimental data and utilize them for the development of unknown substances (materials).
NEXT Eco System
Tweet