E-I Net

E-I Net is a spiking neural circuit model that learns sparse code representations such as found in the brain. E-I Net models aspects of the pattern-learning neural networks found the cerebral cortex by using separate populations of excitatory and inhibitory spiking neurons to find statistical patterns in an input information space. Using only unsupervised learning (i.e. no training or reference signal), E-I Net learns Gabor-like receptive fields such as those found in primary visual cortex (area V1). The spiking circuit simulator is written in MATLAB.

This work was done at the UC Berekeley Redwood Center for Theoretical Neuroscience and is an evolution of SAILnet by Joel Zylberberg.

FAQ: The motivation behind E-I Net is described in the E-I Net FAQ.

Documentation: The inner workings of E-I Net's spiking circuit simulator is explained in this Technical Overview.

Source code: The E-I Net source code and the spiking circuit simulator on which it is based (Neurosim) is written in MATLAB and can be downloaded from GitHub or as a zipfile.


King PD, Zylberberg J, DeWeese MR (2013). Inhibitory interneurons decorrelate excitatory cells to drive sparse code formation in a spiking model of V1. J Neuroscience 33(13): 5475-5485. abstract, PDF

King PD, Zylberberg J, DeWeese MR (2012). Inhibitory interneurons enable sparse code formation in a spiking circuit model of V1. BMC Neuroscience 13(Suppl 1):P148. Presented at Computational Neuroscience Society (CNS) 2012. PDF

Zylberberg J, Murphy JT, DeWeese MR (2011). A sparse coding model with synaptically local plasticity and spiking neurons can account for the diverse shapes of V1 simple cell receptive fields. PLoS Computational Biology 7(10): e1002250. doi:10.1371/journal.pcbi.1002250. article