Abstract: 

There has been huge advancements in the field of Artificial Intelligence in the last decade. While this has been fueled by better compute, more data and cheaper storage, the real success which has revolutionized AI applications is a combination of deep learning and transfer learning. This session will give a gentle introduction to deep learning, deep transfer learning, cover its scope, advantages, applications and limitations. We will then do a deep-dive into three industry focused use-cases pertaining to Semiconductors and Manufacturing where we have leveraged a combination of deep transfer learning and traditional computer vision to solve these challenging, yet interesting problems at nanoscale.

Chip and equipment manufacturing is a tough task given the strict adherence to quality standards and processes like six-sigma control checks. In this session, we will be looking at ways in which we leveraged a combination of traditional image processing and computer vision techniques and coupled it with the power of deep transfer learning and deep learning methodologies including classification, clustering, object detection and text recognition. Following is a brief on the three use-cases we will be covering in the session:
Automated Defect Classification at Nanoscale: EUV masks are critical towards manufacturing of integrated circuit designs. Defects in these masks can be highly costly as they end up degrading the quality of manufactured wafers. Leveraging a combination of traditional image processing, classification and object recognition coupled with deep transfer learning, we will look at how to detect defects at nanoscale from low-resolution scanner images and classify them using a hierarchical multi-level classifier.
Automated Defect Clustering at Nanoscale: Masks are critical to ensure the right patterns are formed on wafer during the manufacturing processed. However there are a wide variety of defects which could occur in these masks. These defects need to be detected and analyzed by process engineers. Classification works well when you know what defects to look at, but how do you deal with unknowns? This is where we leveraged a combination of deep transfer learning coupled with unsupervised learning to build an automated defect clustering solution
Generic Optical Character Recognition for Inventory Tracking: Typically there are a wide variety of inventory equipment which needs to be tracked based on artifacts like barcodes, engraved text, printed text and so on. Usually these artifacts vary across equipment in the form of various formats, shapes, sizes, orientation and textures. We will talk about a generic optical character recognition system which leverages traditional image processing techniques as well as deep learning based object detection models to extract text from equipment with varying formats.
This is going to be a practitioner-focused talk so we will be looking at how deep transfer learning coupled with traditional computer vision has been leveraged for solving the above use-cases and also discuss on model architectures, tools and techniques which were used in the process.

Bio: 

Dipanjan (DJ) Sarkar is a Data Science Lead at Applied Materials, leading advanced analytics efforts around computer vision, natural language processing and deep learning. He is also a Google Developer Expert in Machine Learning. He has consulted and worked with several startups as well as Fortune 500 companies like Intel and Open Source organizations like Red Hat. He primarily works on leveraging data science, machine learning and deep learning to build large- scale intelligent systems. He holds a master of technology degree with specializations in Data Science and Software Engineering.

Dipanjan has been an analytics practitioner for several years now, specializing in machine learning, natural language processing, computer vision and deep learning. Having a passion for data science and education, he also acts as an AI Consultant and Mentor at various organizations like Springboard, where he helps people build their skills on areas like Data Science and Machine Learning. Dipanjan is also a published author, having authored several books on R, Python, Machine Learning, Social Media Analytics, Natural Language Processing, and Deep Learning. In his spare time he loves reading, gaming, watching popular sitcoms and football and writing interesting articles on https://medium.com/@dipanzan.sarkar and https://www.linkedin.com/in/dipanzan. He is also a strong supporter of open-source and publishes his code and analyses from his books and articles on GitHub at https://github.com/dipanjanS.