Units:

12

Description:

This course focuses on the investigation of biologically inspired neural networks that mimic both the functional behavior as well as organizational structure of the mammalian neocortex, with the objective of engineering silicon-based implementations possessing brain-like capabilities. This course introduces a taxonomy of Neural Networks (NN), based on: 1) Neuron Model, 2) Neural Coding, and 3) Learning Paradigm. The taxonomy includes three major classes of NNs: Artificial Neural Networks (ANN), Spiking Neural Networks (SNN) and Temporal Neural Networks (TNN). This course focuses mainly on TNNs because of their "neuromorphic" attributes that strongly adhere to biological plausibility. Based on TNNs, this course explores the potential of designing sensory signal processors, or NeuroMorphic Processing Units (NMPUs), that exhibit both brain-like processing performance and brain-like energy efficiency. In this course, students will work in small teams to pursue semester-long research projects on exploring and prototyping special-purpose NMPUs. These team projects will span applications, architectures, microarchitecture designs, and CMOS implementations of NMPUs. Each team will have the opportunity to define and carry out a specific project with support and guidance from the teaching staff and industry advisors. The team projects can help prepare students for joining industry innovation teams on AI/ML accelerators or academia teams on neuromorphic computing research.

Registration in the course is by invitation only: Student's must send their resume to instructors who will extend invitations to students based on their qualifications and experience.

Prerequisites: 18-740

Last Modified: 2024-11-22 9:33AM

Semesters offered:

• Spring 2025
• Spring 2024
• Spring 2023
• Spring 2022
• Fall 2018
• Fall 2017
• Fall 2016
• Fall 2015
• Fall 2014
• Fall 2013
• Fall 2012
• Fall 2011
• Fall 2010
• Fall 2009
• Spring 2006
• Spring 2005
• Fall 2004
• Fall 2003
• Fall 2002
• Fall 2001