20110124

What is a tempotron? I do not know.

Robert Gütig & Haim Sompolinsky, The tempotron: a neuron that learns spike timing–based decisions, Nature Neuroscience 9, 420 - 428 (2006) doi:10.1038/nn1643

Hey guys, Is anyone familiar with the ideas in this paper? It seems like its some kind of alternative neural network model that tries to explicitly incorporate the idea of spike trains, rather than just being an integrator composed with a nonlinear function at every neuron.

I'm going to try to read it tonight (possibly not in great depth), there is a talk I want to go to:
http://www.math.ias.edu/seminars/abstract?event=40651
. If people are interested I can try to give some kind of summary.


2 comments:

  1. Hello,

    Can you give a summary? source code?

    ReplyDelete
  2. Hi Hananel, I'm afriad Chris posted that to our little site asking for more information -- and none of us had any. This paper is not the work of anyone on Wealone. In general, I have difficulty relating most neural network models to the "real world" -- they are often too abstract to provide practical insight in neuroscience, and unless they offer an advantage over existing machine learning algorithms, are difficult to apply in computer science. But.. what do I know.

    ReplyDelete