Confirmed Talks

Programme (Half-day - Friday morning)

Time Speaker/Company Title
09:00-09:30 Prof. Ahmed Hemani (School of EECS, Sweden) Synchoros VLSI Design for Neuromorphic Applications to achieve 3-4 orders more efficiency than GPUs/FPGAs
09:30-10:00 Prof. Muhammad Shafique (University of Technology, Vienna) Security for Machine Learning Systems: Attacks, Defenses, and Reconfigurability
10:00-10:30 Dr. Stylianos I. Venieris (Samsung AI Center, Cambridge) Current role and future potential of FPGAs in the development of AI-powered consumer-electronics products
10:30-11:00 Break
11:00-11:30 Prof. Nachiket Kapre (University of Waterloo, Canada) Machine Learning on FPGAs: Application + Architecture + CAD
11:30-12:00 Cathal McCabe (Xilinx) Xilinx Machine Learning solutions from cloud to edge
12:00-12:30 Panel Discussion

Programme Overview

    Prof. Ahmed Hemani
  • 09:00-09:30 : Prof. Ahmed Hemani - Dept. of Electronics, School of EECS, KTH, Stockholm, Sweden

    Title:

    Synchoros VLSI Design for Neuromorphic Applications to achieve 3-4 orders more efficiency than GPUs/FPGAs

    Abstract:

    The rise of megatrend in Neural Networks has identified ASIC (custom hardware design) as a necessity to achieve orders of magnitude greater performance and efficiency compared to what is feasible with GPUs and FPGAs. At KTH, we have developed a VLSI Design method called Synchoros VLSI Design Style that enables achieving 3-4 orders better computational efficiency compared to GPUs and FPGAs. The most critical advantage of synchoros VLSI Design is not just its high performance and efficiency but the fact that it can be produced with engineering effort comparable to CUDA programming. Additionally, synchoros VLSI Design Style also promises to dramatically lower the manufacturing cost as well. This talk will introduce this design style and show how it has been applied to different types of Neural Networks.

    Bio:

    Prof. Ahmed Hemani is Professor in Electronic Systems Design at School of ICT, KTH, Kista, Sweden. His current areas of research interests are massively parallel architectures and design methods and their applications to scientific computing and autonomous embedded systems inspired by brain. In past he has contributed to high-level synthesis – his doctoral thesis was the basis for the first high-level synthesis product introduced by Cadence called visual architect. He has also pioneered the Networks-on-chip concept and has contributed to clocking and low power architectures and design methods. He has extensively worked in industry including National Semiconductors, ABB, Ericsson, Philips Semiconductors, Newlogic. He has also been part of three start-ups.

  • Prof. Muhammad Shafique
  • 09:30-10:00 : Prof. Muhammad Shafique - Computer Architecture and Robust Energy-Efficient Technologies, Embedded Computing Systems Group, Institute of Computer Engineering, Vienna University of Technology

    Title:

    Security for Machine Learning Systems: Attacks, Defenses, and Reconfigurability

    Abstract:

    Access to massive amounts of data and high-end computers has heralded revolutionary advances in Machine Learning (ML) impacting domains ranging from autonomous driving and robotics, to healthcare, the natural sciences, the arts and beyond. As we deploy modern ML systems in safety- and health-care applications, however, it is important to ensure their security against adversarial attacks. Researchers have shown that many modern ML algorithms, especially the ones based on the deep neural networks (DNNs) are fragile and can be embarrassingly easy to fool. This is easier said than done. Recent research has shown that DNNs are susceptible to a range of attacks including adversarial input perturbations, backdoors, Trojans, and fault attacks. This can create catastrophic effects for various safety-critical applications like automotive, healthcare, etc. For instance, self-driving cars and vehicular networks, which heavily rely on ML-based functions, exhibit a wide attack surface that can be exploited by well-known and yet-unknown-but-possible attacks on ML models. DNNs contain hundreds of millions of parameters and are hard to interpret/debug let alone verify, significantly increasing the chance they may misbehave. Further, any ML system is only as robust as the data on which we train it on. If the data distributions change in the field, this can impair performance (for example, an autonomous vehicle trained in day time conditions may not function at nighttime).

    The goal of this talk is to shed light on various security threats for the ML algorithms, especially the deep neural networks (DNNs). Various security attacks and defenses for DNNs will be presented. Afterwards, open research problem and perspectives will be briefly discussed. Towards the end, this talk will also highlight the need for reconfigurability in ML systems as a potential means to introduce randomness that can make attackers’ life tough.

    Bio:

    Prof. Muhammad Shafique is a full professor (Univ.Prof.) of Computer Architecture and Robust Energy-Efficient Technologies (CARE-Tech.) at the Embedded Computing Systems Group, Institute of Computer Engineering, Faculty of Informatics, Vienna University of Technology (TU Wien) since Nov. 2016. He received his Ph.D. in Computer Science from Karlsruhe Institute of Technology (KIT), Germany in Jan.2011. Afterwards, he established and led a highly recognized research group for several years as well as 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 ICT R&D project. He has also established strong research ties with multiple universities in Pakistan, where he is actively co-supervising various R&D activities, resulting in top-quality research outcome and scientific publications. Before, he was with Streaming Networks Pvt. Ltd. (Islamabad office) where he was involved in research and development of video coding systems several years.

  • Stelios Venieris
  • 10:00-10:30 : Dr. Stylianos I. Venieris - Researcher at Samsung AI Center, Cambridge

    Title:

    Current role and future potential of FPGAs in the development of AI-powered consumer-electronics products

    Abstract:

    The predictive power of deep-learning models has enabled a wide range of revolutionary user-facing applications. To provide high quality of experience (QoE) while serving a substantially large pool of users in real-world products, dedicated efficiency-boosting techniques are required, from model compression to highly-optimised software and hardware implementations. In this space, the resilience of FPGAs can become a key factor in enabling the design of higher quality AI systems that meet the demanding performance requirements of diverse systems, from ubiquitous mobile devices to complex robot platforms. This talk will introduce the current role of FPGAs in the development of real-world AI systems and present the emerging challenges that are still open to be addressed by reconfigurable computing.

    Bio:

    Dr. Stylianos Venieris is a Researcher at Samsung AI Center-Cambridge in UK. He received his PhD in Reconfigurable Hardware and Deep Learning from Imperial College London in 2018 and his MEng degree in Electrical and Electronic Engineering from Imperial College London in 2014. His research interests include methodologies for the principled and automated mapping of deep learning algorithms on mobile and embedded computing platforms, as well as the design of custom hardware accelerators for the high-performance, energy-efficient deployment of deep neural networks.

  • Nachiket Kapre
  • 11:00-11:30 : Prof. Nachiket Kapre - Electrical and Computer Engineering, University of Waterloo, Canada

    Title:

    Machine Learning on FPGAs: Application + Architecture + CAD

    Abstract:

    As machine learning gets more popular and prevalent, academic research needs to innovate in different, creative ways to stay competitive. In this talk I will review three core contributions of the WatCAG group from the pas few years in Applications, Architectures, and CAD. (1) CaffePresso is a framework for targeting memory-constrained embedded SoC, (2) FPGA cascades-aware, high-frequency overlays for Convolutional networks, and (3) Optimization formulation to explore throughput-latency tradeoffs when mapping to such high-performance overlays.

    Bio:

    Prof. Nachiket Kapre is an assistant Professor within the Department of Electrical and Computer Engineering, University of Waterloo. His research interests include Digital systems, Embedded computing systems and Reconfigurable computing

  • Nachiket Kapre
  • 11:30-12:00 : Cathal McCabe - Xilinx University Program Manager EMEA

    Title:

    Xilinx Machine Learning solutions from cloud to edge

    Abstract:

    It is predicted that worldwide spending on cognitive and Artificial Intelligence systems will reach $77.6B in 2022 adding almost $4T in business value. Demand for more complex systems with varied workloads and different design requirements presents key challenges. The Xilinx ML-Suite is a collection of tools for designing customised Machine Learning networks. The new 7nm Versal ACAP features scalar processing engines, adaptable hardware engines, and software programmable Intelligent engines. ML-Suite along with Versal offer new ways of designing more efficient and higher performance Machine Learning systems from the edge to the cloud.

    Bio:

    Cathal is based in Xilinx Ireland in Dublin. He manages the Xilinx University Program in EMEA and is responsible for supporting universities, developing and delivering training on the latest Xilinx tools and technologies, industrial-academic partnerships, and special initiatives. Cathal is also responsible for the XUP donation program which provides universities with access to Xilinx tools, IP, and hardware platforms for teaching and research.

  • 12:00-12:30: Panel Discussion

  • Panelists:
    • Prof. Ahmed Hemani
    • Prof. Muhammad Shafique
    • Prof. Nachiket Kapre
    • Dr. Stylianos I. Venieris
    • Cathal McCabe