EECS Seminar: Machine Learning at the Edge - Hardware, Software Approaches for Energy Efficiency and Robustness

Abstract: Gigantic rates of data production in the era of big data, Internet of Things (IoT) and smart cyber physical systems (CPS) pose incessantly escalating demands for massive data processing, storage and transmission while continuously interacting with the physical world under unpredictable, harsh and energy/power-constrained scenarios. Therefore, such systems need to support not only the high-performance capabilities at tight power/energy envelop, but also need to be intelligent, cognitive and robust. This has given rise to a new age of machine learning (and, in general artificial intelligence) at different levels of the computing stack, ranging from edge and fog to the cloud. In particular, deep neural networks (DNNs) have shown a tremendous improvement over the past six to eight years to achieve a significantly high accuracy for a certain set of tasks, like image classification, object detection and natural language processing. However, these DNNs require highly complex computations, costing a huge amount of processing, memory and energy budgets. To some extent, Moore’s Law helps by packing more transistors in the chip. However, at the same time, every new generation of device technology faces new issues and challenges in terms of energy efficiency, power density and diverse reliability threats. These technological issues and the escalating challenges posed by the new generation of IoT and CPS systems force us to rethink the computing foundations, architectures and system software. Moreover, in the era of growing cybersecurity threats, the intelligent features of a smart CPS and IoT system face new types of attacks, requiring new design principles for enabling robust machine learning.

In my research group, we are investigating the foundations for the next-generation energy-efficient and robust AI computing systems while addressing the above mentioned challenges across the hardware and software stacks. In this talk, I will present different design challenges for building highly energy-efficient and robust machine learning systems for the edge, covering both the efficient software and hardware designs. After presenting a quick overview of the design challenges, I will present the research roadmap and results from our brain-inspired xomputing (BrISC) project, ranging from neural processing with specialized machine learning hardware to efficient neural architecture search algorithms, covering both fundamental and technological challenges, which enable new opportunities for improving the area, power/energy and performance efficiency of systems by orders of magnitude. Toward the end, I will provide a quick overview of different reliability and security aspects of the machine learning systems deployed in smart CPS and IoT, specifically at the edge. This talk will pitch that a cross-layer design flow for machine learning (and in particular for DNNs), that jointly leverages efficient optimizations at different software and hardware layers, is a crucial step toward enabling the wide-scale deployment of DNNs in resource-constrained embedded AI systems like UAVs, autonomous vehicles, robotics, IoT-health care/wearables, industrial IoT, etc.

Bio: Muhammad Shafique received a doctorate in computer science from the Karlsruhe Institute of Technology (KIT), Germany, in 2011. Afterward, he established and led a highly recognized research group at KIT and conducted impactful research and development activities in Pakistan. Besides co-founding a technology startup in Pakistan, he was also an initiator and team lead of an information and computer technology research and development project. He has also established strong research ties with multiple universities in Pakistan, where he is actively co-supervising various research activities, resulting in top-quality outcomes and scientific publications. Before, he was with Streaming Networks Pvt. Ltd., where he was involved in research and development of video coding systems. In October 2016, he joined the Institute of Computer Engineering at the Faculty of Informatics, Technische Universität Wien, Vienna, Austria, as a professor of computer architecture and robust, energy-efficient technologies. Since September 2020, he is with the Division of Engineering, New York University Abu Dhabi (NYUAD), United Arab Emirates.

Shafique has demonstrated success in leading team projects, meeting deadlines for demonstrations, motivating team members to peak performance levels and completion of challenging tasks. His experience is complemented by strong technical knowledge and an educational record. He also possesses an in-depth understanding of various video coding standards (HEVC, H.264, MVC, MPEG-1/2/4). His research interests are in brain-inspired computing, artificial intelligence and machine learning hardware and system-level design, energy-efficient systems, robust computing, hardware security, emerging technologies, FPGAs, MPSoCs and embedded systems. His research has a special focus on cross-layer analysis, modeling, design and optimization of computing and memory systems. The researched technologies and tools are deployed in application use cases from Internet-of-Things, smart cyber-physical systems and ICT for development domains.

Shafique has given several keynote and invited talks and tutorials at premier venues. He has also organized many special sessions at premier conferences (DAC, ICCAD, DATE, IOLTS and ESWeek) and served as the guest editor for IEEE Design and Test Magazine and IEEE Transactions on Sustainable Computing. He has served as the TPC chair of ISVLSI, PARMA-DITAM, RTML, ESTIMedia and LPDC, general chair of DDECS and ESTIMedia, track chair at DATE, IOLTS, DSD and FDL, and Ph.D. forum chair of ISVLSI. He has also served on the program committees of numerous prestigious IEEE/ACM conferences, including ICCAD, ISCA, DATE, CASES, ASPDAC and FPL. He is a senior member of the IEEE and IEEE Signal Processing Society and a member of the ACM, SIGARCH, SIGDA, SIGBED and HIPEAC. He holds one U.S. patent and has co-authored six books, 10-plus book chapters, and over 250 papers in premier journals and conferences.

Shafique received the prestigious 2015 ACM/SIGDA Outstanding New Faculty Award, the 2020 AI-2000 Chip Technology Most Influential Scholar Award, six gold medals in his educational career, and several best paper awards and nominations at prestigious conferences like CODES+ISSS, DATE, DAC and ICCAD, Best Master Thesis Award, DAC'14 Designer Track Best Poster Award, IEEE Transactions of Computer "Feature Paper of the Month" Awards, and Best Lecturer Award. His research work on aging optimization for GPUs was featured as a research highlight in Nature Electronics, February 2018 issue.

 Host: Mohammad Al Faruque