2011/nengo11

Nengo and Large-Scale Neural Modelling

Members: Aleksandrs Ecins, Chris Eliasmith, David Noelle, Dimitris Pinotsis, Merve Kaya, Fabio Stefanini, Francesco Galluppi, Garrick Orchard, Je Hi An, Roi Kliper, Kwabena Boahen, Katie Zhuang, Mounya Elhilali, Michael Pfeiffer, Ravi Shekhar, Runchun Wang, Samantha Adams, Shih-Chii Liu, Samuel Shapero, Timir Datta, Tomas Figliolia, Timmer Horiuchi, Troy Lau, Andre van Schaik

This tutorial is meant to give a hands-on introduction to Nengo, a neural modelling toolkit. The main difference between Nengo and other neural modelling approaches is that we can directly specify a particular computation to be performed by the synaptic connections from one neural group to another. For example, if one neural group of 100 neurons represents velocity (a 2-dimensional vector) and we want to have another neural group of 100 neurons represent position (the integral of velocity), Nengo can solve for the feed-forward and recurrent 100x100 connection weight matrices that will compute that function.

Nengo is written primarily in Java, with a Python scripting interface. The latest version is available at  http://ctn.uwaterloo.ca/~cnrglab/f/nengo-1.3b.zip and it runs on Linux, OS X, and Windows. To install, unzip the file and run nengo (under Linux and and OS X) or nengo.bat (under Windows).

Tutorial Material