INEMI Smart Manufacturing Tech Topic Series: Enhancing Yield and Quality with Explainable AI

Описание: A Deep Topological and Self-Supervised Learning Approach
Janhavi Giri, PhD (Intel Foundry Automation)
May 6, 2025

In this webinar, Janhavi Giri, PhD, (Intel Foundry Automation) introduces an explainable AI framework that integrates deep topological data analysis (DTDA) and self-supervised learning (SSL). This novel approach autonomously extracts patterns and detects defects from vast and complex semiconductor manufacturing image data without requiring labels. Data is key and the DataRefiner© platform discussed by Dr. Giri provides a way to manage data for improved yield and reliability while supporting industry’s goals in systems integration and sustainability. Her presentation shows how:

• DTDA and SSL are extremely beneficial for image segmentation and detecting unique patterns
without requiring any labeled data.
• Leveraging the capability enabled in the DataRefiner© platform:
• Provides unsupervised image segmentation that enables engineers to quickly analyze, visualize,
and identify categories/sub-categories for their unlabeled image datasets.
• Significantly reduces the learning time generating desired data separation with transfer learning
•. Enhances engineering productivity, yield, quality, and reduces costs.

Download a copy of the webinar presentation:
https://thor.inemi.org/webdownload//2...