Machine Learning Technique Sharpens Prediction of Material’s Mechanical Properties

Original article can be found here (source): Communications of the ACM – Artificial Intelligence

By Nanyang Technological University (Singapore)
March 26, 2020
Comments

An illustration of the technique.

Scientists at Nanyang Technological University, Singapore, the Massachusetts Institute of Technology, and Brown University have developed new approaches that significantly improve the accuracy of a material testing technique by harnessing the power of machine learning.

Credit: MIT

Scientists at Nanyang Technological University, Singapore (NTU Singapore), the Massachusetts Institute of Technology, and Brown University have applied machine learning to enhance the accuracy of material testing.

The scientists developed and trained a neural network to predict material samples’ yield strength with 20 times greater accuracy than current methods, by feeding it experimentally-measured data from a standard nano-indentation process, in combination with synthetic data.

This hybrid multi-fidelity method tapped deep learning algorithms.

Said NTU Singapore’s Subra Suresh, “The use of real experimental data points helps to compensate for the ideal world that is assumed in the synthetic data. By using a good mix of data points from the idealized and real world, the end result is drastically reduced error.”

From Nanyang Technological University (Singapore)
View Full Article

 

Abstracts Copyright © 2020 SmithBucklin, Washington, DC, USA


 

No entries found