Friday, June 22, 2018 – Machine Learning Today and Tomorrow: Technology, Circuits and System View

Brief description: A wide ranging full day, focusing on how technology and circuit designers engage in and drive the future of AI/ML systems.

Topical content:

  • Foundations and performance metrics for ML systems
  • Advanced and emerging circuit architectures and technologies for next-generation ML systems
  • Tools and datasets for benchmarking and evaluation of SoA ML systems

Emerging and future applications for ML systems, driving circuits and devices researchAbstract: Machine learning has emerged and continues to evolve as an extremely rich and impactful driver of integrated systems. Given the rapid progression both in algorithmic complexity and application platforms, all signs are that technology and circuit designers will play a critical role in maximizing the impact today and in the future that machine-learning systems can have. To become proactive drivers of the advances that will be made in this area, these communities must understand the complex and dynamic landscape of machine-learning algorithms, their uses, and the technological trends driving their progression. This session aims to provide this background, as well as the preparation needed to engage in the future of machine-learning systems.


8:00a Open
8:50a What different forms of machine learning exist, both within and beyond neural networks, and what are the different problems they solve. (at what tasks does ML struggle today?) Vivienne Sze (MIT)
How do we evaluate ML, in terms of the different metrics and their tradeoffs David Brooks (Harvard)
9:30a Joint Q&A Speakers 1&2
9:50a Break (20min)
10:10a How do people use ML learning today, in terms of tools and SW packages (what issues are faced in terms of public safety and liability) Marc Duranton (CEA-LETI)
10:45a Unconventional & mixed-signal circuit architectures for ML Boris Murmann (Stanford)
11:20a Joint Q&A Speakers 3&4
11:40a Lunch (30min)
12:10p Lunch panel: How VLSI circuit and technology experts will drive the machine-learning revolution.B. de Salvo (CEA-LETI, J. Duguchi (Toshiba), N. Shanbhag (UIUC)
1:40p ML circuits for memory-centric computingNaresh Shanbhag (UIUC)
2:15pML technologies for memory-centric computingElisa Vianello (CEA-LETI)
2:50p Joint Q&A Speakers 5&6
3:10p Break (20min)
Emerging applications: humanoid robotics Tetsuya Ogata (Waseda Univ )
4:05p Emerging applications: ML in the IoTDaisuke Okanohara (Preferred Net.)
4:40p Joint Q&A Speakers 7&8
5:00p Close

Satellite Workshop

2018 Silicon Nanoelectronics Workshop

will be co-located with the Symposia on Sunday and Monday, June 17-18, 2018 at the Hilton Hawaiian Village.

2018 Spintronics Workshop on LSI 

will be co-located with the Symposia at a date to be announced.