Cadence: Bullish on AI

Estimated reading time: 2 minutes

David White has been involved with artificial intelligence research for almost 30 years. Now, David is the senior group director of R&D for Cadence Design Systems, and I knew we’d have to speak with him for this month’s issue on AI. In a recent interview, we discussed his decades of work in AI, Cadence’s research into AI and machine learning, and what he believes AI could mean for the EDA tools of the future.

Andy Shaughnessy: Tell us a little about your background, your work with AI, and your thoughts on AI overall.

David White: I started working in AI in 1989 as a college student after discovering a copy of Parallel Distributed Processing, by David Rumelhart. I was so enthralled that I completed my undergraduate thesis on using neural network-based controls for a robotic arm. That work led me to McDonnell Douglas, now Boeing, where I worked in the New Aircraft Products Division on machine learning research for manufacturing and flight controls. As a result of this work, NSF asked me to chair the first NSF Workshop on Aerospace Applications of Neural Networks, which included machine learning researchers from across the country as well as a presidential science advisor and government officials.

I joined the MIT AI Laboratory where I continued my research and edited and co-authored a book on intelligent decision and control systems in 1992 with leaders in the machine learning world such as Michael Jordon, Paul Werbos and Andy Barto. I completed my graduate work at MIT where my research applied machine learning and chemometrics to semiconductor processing. I later co-founded and served as CTO of Praesagus, a company that was acquired by Cadence in 2006, and I have been working on electronic design automation with Virtuoso technology since 2009.

In terms of my thoughts on AI, I am really excited about the prospects of building intelligent decision systems that can learn from users and their environment. We believe we are bringing a unique perspective to how we build these systems. We are combining innovations in machine and deep learning with large scale optimization and distributed processing in unique ways. Much of what we are working on has applications beyond EDA and extends to how we can build design and analysis software that tailors itself to the user and their mission.

Shaughnessy: How did Cadence first get involved with AI?

White: I joined Cadence in 2006 when our company was acquired, so my frame of reference begins then. Cadence’s research in machine learning (ML) for physical design and electrical analysis started in the 2009-2010 timeframe, with two persons and myself. Our motivation came from observing the scale and complexity that grew with the increase in data such as larger designs, larger simulations, etc.

To address these problems, we began to look at data-driven solutions such as analytics and machine learning. When we began the work, there was not the same buzz around machine or deep learning, and we just found it to be a useful tool to create fast models of complex non-linear problems that required long compute times using more traditional methods.

To read this entire interview, which appeared in the September 2018 issue of Design007 Magazine, click here.