Communications of the ACM
Machine Learning Models Can Produce Reliable Results with Limited Training Data
The researchers found that PDEs that model diffusion have a structure that is useful for designing AI models.
Credit: Getty Images
Researchers at the U.K.'s University of Cambridge and Cornell University demonstrated that machine learning models can generate reliable results even with limited training data.
The researchers focused on partial differential equations (PDEs), which are considered the building blocks of physics.
Cambridge's Nicolas Boullé explained, "Using a simple model, you might be able to enforce some of the physics that you already know into the training dataset to get better accuracy and performance."
In developing an algorithm for predicting the solutions of PDEs under various conditions, the researchers exploited the short and long-range interactions of PDEs to build mathematical guarantees into the model and calculate the amount of training data needed to ensure reliability.
Said Boullé, "It's surprising how little data you need to end up with a reliable model. Thanks to the mathematics of these equations, we can exploit their structure to make the models more efficient."
From University of Cambridge (U.K.)
View Full Article
Abstracts Copyright © 2023 SmithBucklin, Washington, D.C., USA
No entries found