Nanotronics uses artificial intelligence to give our customers’ unprecedented freedom and control for defect detection. We offer an AI based Anomaly Detection Algorithm (ADA) toolkit that automates the work of writing computer vision algorithms to detect and classify defects on bare substrate and epi wafers as well as on thin films, glass and any other material with a uniform background.
Our architecture uses customized enhancements to deep learning and convolutional networks geared towards sparse data, along with specialized detection algorithms, specifically tuned for speed and high-quality inference from sparse data. All aspects of the architecture have been engineered to utilize the parallel power of modern GPUs.
What this means in practice: our customers have successfully trained ADA to detect defects with as little as ten sample images, (thirty on average) and then import the algorithm into a production tool in order to begin classifying a new defect type.
In Q1 2018 we will be releasing a general artificial intelligence solution for patterned wafers and devices that will require virtually zero training data. To schedule an early demo of this software please email our sales team.