Sorry, you need to enable JavaScript to visit this website.

AI at the Edge Workshop in IGSC

Dr. Merkel is organizing a workshop on AI at the Edge to be held in conjunction with the 2020 International Conference on Green and Sustainable Computing (

Artificial intelligence (AI) has established itself as a valuable tool to improve multiple aspects of human life.  Today, most AI applications leverage power-hungry cloud resources (e.g. GPUs) in order to meet their ever-growing memory and compute requirements.  While moving AI to the edge will unlock many new application domains, it remains very challenging to perform the required computations on size, weight, and power (SWaP)-constrained devices (mobile phones, sensors, etc.).  To help address the challenge, this workshop will bring together experts in multiple fields such as AI, integrated circuit design, device physics, and computer architecture to discuss the way forward in moving AI onto SWaP-constrained edge devices.  This workshop will complement a number of this year’s IGSC topics/themes including VLSI design, computer architecture, embedded design, low-power design, and ML for energy-efficient system design.

Topics include but are not limited to:

  • AI algorithm efficiency (quantization, compression, etc.)
  • Energy-efficient AI architectures
  • AI circuit optimization for size, weight, and power (SWaP) efficiency
  • Neuromorphic computing (based on conventional or emerging technologies)
  • Efficient (e.g. energy-efficient) edge-cloud interaction for AI applications
  • Brain-inspired machine learning algorithms


Contributions are limited to 8 pages (IEEE style) and will be published in IEEE Xplore.  Papers can be submitted through Easy Chair: 

Important Dates:

August 15, 2020 September 5, 2020 First submission due.

September 1, 2020 September 21, 2020 Final camera-ready submission due.

For quesions, please contact Cory Merkel:

Workshop Agenda - Monday, October 19 (All times are EST)

8:00-8:05 am - Welcome by Cory Merkel

Session 1 (Chair:  Cory Merkel)

  • 8:05-8:30 am - "Low Size, Weight, and Power Neuromorphic Computing to Improve Combustion Engine Efficiency" by Catherine Schuman, Steven Young, John Mitchell, Travis Johnston, Derek Rose, Bryan Maldonado and Brian Kaul
  • 8:30-8:55 am - "Conversion of an Unsupervised Anomaly Detection System to Spiking Neural Network for Car Hacking Identification" by Yassine Jaoudi, Chris Yakopcic and Tarek Taha
  • 8:55-9:20 am - "Role of Memristive Device Variability in Governing the Accuracy and Conversion Time of Machine Learning Algorithms on Neuromorphic Hardware" by Andrew Ford and Rashmi Jha
  • 9:20-9:45 am - "Memristor Based Neuromorphic Network Security System Capable of Online Incremental Learning and Anomaly Detection" Md Shahanur Alam, Chris Yakopcic, Guru Subramanyam and Tarek M Taha

9:45-10:00 am - Break

Session 2 (Chair: TBD)

  • 10:00-10:25 am - "Towards Green DNNs for the Edge with Tapered Fixed-Point" by Vedant Karia, Hamed Fatemi Langroudi and Dhireesha Kudithipudi
  • 10:25-10:50 am - "Dynamic Energy-Accuracy Scaling in AI Hardware" by Cory Merkel
  • 10:50-11:15 am - "Memristive devices and arrays for neuromorphic computing" by J. Joshua Yang