What is it about?
We are using a novel explainable AI method, termed as zero-bias deep learning, for defect detection in the 3D printing process. Not only do we enable the system to detect unknown defects by only learning normal and samples with known defects, more importantly, our approach turn the vanilla deep learning pardigm from a blackbox to an explainable and trustworthy process.
Featured Image
Photo by vackground.com on Unsplash
Why is it important?
We enable the system to detect unknown defects from only learning normal and samples with known defects, but also, our approach turn the vanilla deep learning pardigm from a blackbox to an explainable and trustworthy process.
Perspectives
Our method not only works in this domian, but also works pretty well in the domain of signal transmitter recognition, e.g. we even tested it in ADS-B systems for authenticating aircraft transponders.
Yongxin Liu
Embry-Riddle Aeronautical University
Read the Original
This page is a summary of: Zero-bias deep neural network for defect detection in composite additive manufacturing using multisource in-situ data, January 2024, American Institute of Aeronautics and Astronautics (AIAA),
DOI: 10.2514/6.2024-0264.
You can read the full text:
Resources
Contributors
The following have contributed to this page
