Using Advanced Micro/Nano-electronic Technology to Establish Neuromorphic Systems Panel

Using Advanced Micro/Nano-electronic Technology to Establish Neuromorphic Systems

Chair: Wei Wang
Panelists:

• Prof. Garrett Rose, ECE Department of Polytech University
  grose@duke.poly.edu
• Dr. Vladimir Gorelik, Founder of Neuronix
  v.gorelik@comcast.net
• Dr. Shih-Chii Liu, Institute of Neuroinformatics, Zurich, Switzerland
  shih@ini.phys.ethz.ch
• Prof. Masud Chowdhury, ECE Department, University of Illinois at Chicago
  masudh@gmail.com
• Dr. Mathew Hynd, Wadsworth, Albany, NY

To start the panel, I will raise several questions regarding the challenges of designing and building devices, circuits, and systems for neuromorphic applications. Then, each panelist will speak for 10-20 minutes about their perspectives to these challenges. Then, the audiences can ask questions to the panel.

Some of the questions asked will be:

In the device level, what is current research status using the CMOS-related technology (analog or mixed signal) to build neuromorphic systems? What are the main challenges? Can CMOS-nano hybrid approach really be an efficient and feasible approach and show improvement over the CMOS-based neuromorphic systems? Some panelists are using CMOS, some are using nanodevices. They will have different perspectives.

In the circuit level, should we use simplified synapses/neurons (digital multivalue) or complicated analog circuitries? Some panelists are suggesting to use complicated synapses/neurons that really mimick the human system (analog) but are difficult to build. Some panelists are using a binary or mutlivalue digital circuit to build a simplified synapse/neuron. The panel can discuss these different approaches.

In the system and architecture level, what is the impact of the model, algorithm and architecture of the neuromorphic system impact to the hardware development? Since the real human cortex is a 3D structure, is it necessary to use 3D IC architecture to build such systems?