This course aims to teach learners about neuromorphic chips and how they address the nanostransistor challenge. The learning outcomes include understanding the concept of neuromorphic computing, implementing dendritic computation with subthreshold analog, and comparing digital, analog, and neuromorphic computing. The course covers skills such as approximating dynamic transformations with the Neural Engineering Framework (NEF) and building whole-brain models with spiking neurons. The teaching method involves a seminar format with a focus on theoretical concepts and practical examples. This course is intended for individuals interested in advanced topics in computer science, neuromorphic engineering, and artificial intelligence.
Stanford Seminar - Neuromorphic Chips - Addressing the Nanostransistor Challenge
Stanford University via YouTube
Overview
Syllabus
Introduction.
Acknowledgments.
A dollar bought more and more transistors until 2014.
Beware of potholes!.
Accidents happen!.
Combining digital communication with analog computation.
Digital versus Analog versus Neuromorphic.
Neuromorphs' Five-Point "Secret" Master Plan 1. Implement dendritic computation with subthreshold analog.
Overlapping dendrites eliminate axon's terminal branches.
Hierarchical axonal branching minimizes wiring.
Neurogrid's overlapping dendrites and hierarchical axons.
Neurogrid versus TrueNorth.
What is the "VHDL" neuromorphic chips? 1981: Wiring Gates.
The Neuromorphics Project.
Nengo, Neuromorph, and Brainstorm.
The Neural Engineering Framework (NEF).
NEF vesus Conventional Neural Network.
Approximating dynamic transformations with NEF.
SPAUN: First whole-brain model built with spiking neurons.
Vector rotation example.
Works on a neuromorphic chip!.
Robot-arm controller video.
Taught by
Stanford Online
