Education Class D3


Title: Machine Learning for  Manycore System Design and Optimization

Instructor: Biresh Kumar, Duke University and Jana Doppa, Washington State University

Abstract: Advanced computing systems have long been enablers for breakthroughs in science and engineering applications including Artificial Intelligence (AI) and Machine Learning (ML) either through sheer computational power or form-factor miniaturization. However, as algorithms become more complex and the size of datasets increase, existing computing platforms are no longer sufficient to bridge the gap between algorithmic innovation and hardware design due to nearing the end of Moore’s law.  This educational lecture will focus on how manycore systems designed by leveraging the benefits of emerging technologies (e.g., three-dimensional integration, processing-in-memory) and machine learning have the potential to bridge this growing gap. First, we will provide the basics on manycore systems — compute layer, interconnect layer, and memory layer — and discuss the challenges of data movement. Second, we will discuss the advantages of heterogeneous computing, 3D integration, and processing-in-memory in addressing the data movement challenge. Third, we will explain the challenges of application-specific manycore systems design due to large combinatorial spaces and the need to perform expensive simulations to optimize multiple conflicting objectives (e.g., power, performance, and temperature). Fourth, we will describe how machine learning can enable efficient and accurate manycore systems design optimization. Finally, we will discuss hardware and software co-design methodologies for emerging deep learning applications.

The lecture will be delivered in a way that any undergraduate student from ECE or CS can understand the material. No background on manycore systems or machine learning is assumed.

Bio: Jana Doppa is the George and Joan Berry Distinguished Associate Professor in the School of Electrical Engineering and Computer Science at Washington State University, Pullman. He received his Ph.D. degree in Computer Science from Oregon State University and his M.Tech. degree from Indian Institute of Technology (IIT), Kanpur. His primary research focus is at the intersection of machine learning and electronic design automation by exploring the synergies between these two mutually beneficial areas.

His research has been recognized with a number of awards, honors, and distinctions including the 2019 National Science Foundation CAREER Award; the 2021 Early Career Award in AI by the International Joint Conference on Artificial Intelligence for ML algorithms to accelerate design automation for science and engineering applications including electronic design automation; the 2021 Best Paper Award from ACM Transactions on Design Automation of Electronic Systems; the 2013 Outstanding Paper Award from the AAAI Conference on Artificial Intelligence; the 2018 Best Student Abstract Award from the AAAI Conference on Artificial Intelligence; the 2015 Outstanding PhD Dissertation Award from Oregon State University and was nominated for ACM Doctoral Dissertation Award; a 2015 Google Faculty Research Award; the 2013 Outstanding Graduate Student Award from College of Engineering, Oregon State University; the 2020 Outstanding Junior Faculty Research Award and the 2018 Reid-Miller Teaching Excellence Award from the College of Engineering, Washington State University.

Biresh Kumar Joardar is an NSF sponsored Computing Innovation Fellow (postdoctoral researcher) in Electrical and Computer Engineering at Duke University mentored by Prof. Krishnendu Chakrabarty. He received his Ph.D. degree in Computer Engineering from Washington State University and his BE degree in ECE from Jadavpur University. His primary research is at the intersection of manycore systems design and machine learning with a current focus on reliable ML on unreliable hardware.

His research has been recognized with a number of awards, honors, and distinctions including a 2021 DAAD AInet Fellowship; a 2020 NSF Computing Innovation Fellowship; the 2019 Best Paper Award at NOCS; Nominated for Best Paper Award at DATE-2020 and DATE-2021 conferences; the 2019 Outstanding Graduate Student Researcher Award from the College of Engineering, WSU; and the 2018 Harold and Dianna Frank Electrical Engineering Fellowship, WSU.