CNS/EE/Be182, Winter quarter 2007
Introduction to neuromorphic and bio-inspired mixed-signal VLSI

Tues and Thurs 9-10AM, Room 080 Moore, begins Thursday January 4th, last lecture Thursday March 8th

CNS182 discussion group.

Week of Lecture (Tues 9-10AM) (D=demo) Lecture (Thurs 9-10AM) Reading Assignment Additional
Jan 1   Orgainizing/Admin/Introduction
Neuromorphic Engineering (T,S)
Book chap. 1, Introduction
Mead, Proc. IEEE 1990, "Neuromorphic Electronic Systems "

Assignment 1

Assignment 1 Solutions (Wei Li's)

Analog circuit simulator
Jan 8 Semiconductor physics (T) Basic sub-threshold transistor behavior (T) Book chap. 2, Semiconductor physics
Book chap. 3, MOSFETs
   
Jan 15

Physio Friend Chip (D)
Web page (T)

Assignment 1 due

2-3 FET Ckts (Current mirror, SrcFoll, DiffPair) (S)

Book chap. 5, Static circuits

Physio Friend paper

Assignment 2

Assignment 2 Solutions (again Wei Li's)

 
Jan 22 Transconductance amp, Follower aggregator (S)
Drain conductance, voltage amplifiers, gain-bandwidth (T)
     
Jan 29 Current conveyor, correlator, bump ckt (S)

Phototransduction (T)

Assignment 2 due

Midterm exam out

Book chapter 6

Bump circuit paper - the original paper on the bump and bump-antibump circuits

Delbruck, 1993, "Analog VLSI Phototransduction" (web page) - a long paper with detailed consideration of many aspects of continuous-time photoreceptor circuits

   
Feb 5 Mismatch and compensation techniques (T) Neuron circuit (S)

Liu et al, 2001, "Orientation-selective spiking neurons" - shows how to use arrays of I&F spiking neurons to make orientation selective neurons

Mahowald, Douglas, 1991, "A Silicon Neuron" - this Nature paper shows how to build a spiking circuit that models Na, K and Ca spike mechanism and adaptation

     
Feb 12 WTA circuits (S) Synapse circuits (S)

Lazzaro's NIPS WTA paper - lays out the analysis of the winner take all circuit, including static and dynamic responses and the first "localized" WTA

Lazzaro's Caltech CS Dept Technical Report, WTA - a more detailed version of the NIPS papers with all the computations.

 

 
Feb 19

Generating current biases (T)

Assignment 3 out

Stocker's Optical Flow Sensor (T)

Bult&Geelen, 1992, Current Division Technique - shows a beutiful, consise argument about the generality of the current splitter principle

Delbruck, AICSP 2005 Fixed Biases paper - analyzes Widlar's bootstrapped mirror and combines it with the current splitter for generating a set of fixed biases

Delbruck ISCAS 2006 Programmable Biases paper - extends the fixed biases to generate programmable bias currents

Stocker's 2006 paper in AICSP on Optical Flow Chip

Assignment 3

Assignment 3 solutions (Wei Li's)

Web page for bias generator design kits

Web page for Stocker's Optical Flow Sensor

Feb 26

AER, WTA demo (D) (S)

Assignment 3 due

Assignment 4 out

CSEMs event based spatial contrast vision sensor (T)

Boahen, Sci Am, 2006, "Neuromorphic Microchips" - a recent paper about the Boahen group work on systems that use AER for communication and computation.

CSEM's frame-event-based spatial contrast vision sensor paper (JSSC 2003)

 

Assignment 4

recep.lgf (partial Analog circuit for simulating photoreceptors)

Assignment 4 Solutions (mixed sources)

CSEM's web pages (devise)

 

Mar 5 Silicon Cochlea (D) (S)

INI's temporal contrast retina (D) (T)

Assigment 4 due

Final handed out

INI's event-based event-based temporal contrast vision sensor (ISSCC 2006)

Draft of submitted paper on temporal contrast retina

  INI's temporal contrast vision sensor web pages
Mar 12   Exams due 15 March      
Mar 19 Grades        

Taught by visiting faculty Shih-Chii Liu (shih@ini.phys.ethz.ch) and Tobi Delbruck (tobi@caltech.edu), Inst. of Neuroinformatics, UNI-ETH Zurich

This class is an introduction to neuromorphic and bio inspired hybrid analog/digital VLSI. Topics include the role of neurally-inspired computational architectures in the context of neuroscience and electrical/computer engineering, transistor physics, basic neuromorphic circuits, system architectures for biological sensors (e.g., retina and cochlea), and spiking neural network implementations.

This course derives from Carver Mead's CNS182 course taught from 1988-1997 and the Physics of Computation course taught by Feynman, Hopfield and Mead. These courses led to the Computation and Neural Systems program. In the new CNS182 course, students will learn about the state-of-the-art in neuromorphic electronics. Lectures will include several live demonstrations of neuromorphic electronic systems, and depending on the available infrastructure, the course may include several lab sessions. The 2 lectures each week will cover technical and "cultural" aspects of neuromorphic electronics and will be weighted towards more technical material at the beginning of the quarter.

Course prerequisties: EE10 (basic electronics) or EE114 (analog integrated circuit design) or equivalent are very helpful. CNS100 and/or CNS187 are also helpful for background.

9 units credit (2-0-7). Grading will be based on homework assigments, midterm, and final exam.

The book for this course is Analog VLSI, Circuits and Principles. You can purchase one at a discount in class.

Last updated 08-Mar-2007 15:21 . Report web page problems to tobi@caltech.edu