Skip to content

Independent AI & deep-tech news

Artificial Intelligence

Less data, faster discovery: Researchers rethink AI’s role in developing new materials

Less data, faster discovery: Researchers rethink AI’s role in developing new materials
Image Credit: https://techxplore.com/

Developing new materials often begins with understanding their electronic structure, or how electrons are arranged and behave within them. Calculating these electronic structures usually needs expensive simulations on supercomputers. That’s led researchers to train artificial intelligence models on millions of simulated data points to make faster predictions.

But according to a current study led by associate professor Susanta Ghosh in Michigan Tech’s Department of Mechanical and Aerospace Engineering, that much training data isn’t important.

Collaborating with researchers from the University of California, Los Angeles (UCLA), Ghosh and his team discovered that AI models require a lot less data to make accurate predictions.

In a study lately published in npj Computational Materials, the research team found that AI models should discard 90% of their training data with virtually no loss in accuracy.

Even after removing 99% of the training data, most predictions stayed highly correct, while the models trained 3 or more times faster. The new findings apply to an expansive range of materials, including metals, water and complicated alloys.

How can an AI model make correct electronic predictions with much less data? Ghosh mention the answer is to use a minimal data set carefully selected by researchers to capture the diversity of the data.

“Much of the electronic structure data repeats the same information. A small, well-chosen sample captures almost the whole thing the AI require to learn,” stated Ghosh, an associate professor inside the Department of Mechanical and Aerospace Engineering.

These findings task the necessity of extensive data sets for electronic structure predictions and could make AI-driven materials development faster and much less costly. Ghosh, who is a faculty member of the Center for Artificial Intelligence in Michigan Tech’s Institute of Computing and Cybersystems, believes this could have a large effect on the field of materials development and modeling.

“By identifying representative electronic-structure data and reducing training time via at least threefold, this work can accelerate AI-driven discovery of complex alloys across tremendous composition space,” he stated.

Written by

Tarun Khanna

577 articles on DeepTechBytes