Announcements

GSSE Seminar by David Shan Hill Wong - Tuesday, 21 May 2019, 10:00 Room ENG Z15

Author: Graduate School of Sciences and Engineering
Time: 10:00
Location: ENG Z15

******************************

KOÇ UNIVERSITY

GRADUATE SCHOOL OF SCIENCES AND ENGINEERING

SEMINAR SERIES

******************************

Speaker: David Shan Hill Wong - Department of Chemical Engineering National Tsing Hua University

 

Title: Integrating Data Science into Process System Engineering Experiences

 

Date: 21 May 2019

 

Time: 10:00-11:00

 

Cookie & Tea: 10:00

 

Place: ENG Z15

 

Host: Prof. Yaman Arkun

 

Abstract

            In recent years, keywords like “big data”, “machine learning” and “smart manufacturing” have become extremely trendy in business and industry.   

 

“Big Data” refers to the situation that data comes in velocity, volume and varieties that are beyond human’s capability to analyze.  “Machine Learning” are methodologies of generating information/knowledge (models) from data using computers with minimal human intervention.  A process capable of implementing information/knowledge (models)-based decisions in order to adapt to external or internal changes is a “Smart Manufacturing” process.

 

However, different manufacturing processes have different characteristics. For example, a refinery is essentially a continuous process that has to deal with seasonal changes in demand mix, supply of crudes.  A polymer plant may consist of a mix of batch and continuous operations that has to perform frequent grade changes.  A semiconductor foundry produces many different products for different customers by a manufacturing line consisting of many steps, each of which is a batch operation performed by many similar tools.  For such a process, changes are norm rather than exception.

 

In this presentation, we shall go through a semi-chronological narratives of our research experiences in leveraging data analytics to augment intelligence in design/optimization, control and monitoring of semiconductor manufacturing and traditional bulk chemical/refining processes.

 

            While data analytics may be powerful, there are always the concerns that: information embedded in the data may not be enough for capturing the true physics of the process, and that data may come in many forms other than numerical values.   Hence, we shall also report some of our recent attempts on how to integrate existing knowledge with data science and leverage the ability of deep learning to handle data of various structures. 

 

Biography: Please see attached.