Spike train feature extraction by Bayesian binning 0.1

Dominik Endres — Wed, 24 Sep 2008 16:19:14 -0000

binsdfc is a command line implementation of the algorithm described in Endres,Oram,Schindelin,Foldiak:Bayesian binning beats approximate alternatives: estimating peri-stimulus time histograms, Advances in NIPS, 2007. It computes spike density functions (SDF) or peri-stimulus time histograms (PSTH). Given that it performs exact Bayesian averaging, the result is somewhere in the middle between the two: while the underlying model is comprised of bins like a PSTH, the averaging process gives rise to a more "continuous" prediction like a SDF. binsdfc accepts input from stdin or a supplied file, and computes the expected SDF, its variance, latencies (as described in Endres, Schindelin, Foldiak, Oram (2007): Examining the joint neural code of latency and firing rate by Bayesian binning, paper will follow soon), and various other averages such as marginal likelihoods and model complexity posteriors. This work is supported by a MRC training fellowship.

mloss.org Spike train feature extraction by Bayesian binning