Presentation 2009-01-20
Which model can properly describe dynamics and smoothness of firing rate?
Ken TAKIYAMA, Kentaro KATAHIRA, Masato OKADA,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) We construct the algorithm using belief propagation(BP), which algorithm simultaneously estimates firing rate and calculates marginal likelihood. Our algorithm can determine the degree and the form of the firing rate smoothness based on hyperparameter estimation and model selection by maximizing the marginal likelihood. Prior distribution is Line process model, Gauss model, or Cauchy model. We discuss which model is appropriate to describe the firing rate and whether appropriate model changes or not depending on the functional form of the firing rate. We conduct two firing rate estimation experiments: the first firing rate evolves smoothly, but the second firing rate involves discontinuity. The second experiment is assumed that the extrinsic stimuli whose timings are entirely-unknown are added to the neuron. The Line process model shows the largest mariginal likelihood value of the three models in both experiments. The Line process model also being able to estimate the unknown stimulus timings, we show the effectivness of the Line process model in the firing estimation issue.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Firing rate estimation / Belief Propagation / Bayesian estimation / Line process / Gaussian graphical model / Hyperparameter estimation / Model selection
Paper # NC2008-98
Date of Issue

Conference Information
Committee NC
Conference Date 2009/1/12(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Neurocomputing (NC)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Which model can properly describe dynamics and smoothness of firing rate?
Sub Title (in English)
Keyword(1) Firing rate estimation
Keyword(2) Belief Propagation
Keyword(3) Bayesian estimation
Keyword(4) Line process
Keyword(5) Gaussian graphical model
Keyword(6) Hyperparameter estimation
Keyword(7) Model selection
1st Author's Name Ken TAKIYAMA
1st Author's Affiliation Graduate School of Frontier Sciences, The University of Tokyo()
2nd Author's Name Kentaro KATAHIRA
2nd Author's Affiliation Graduate School of Frontier Sciences, The University of Tokyo:Brain Science Institute, RIKEN
3rd Author's Name Masato OKADA
3rd Author's Affiliation Graduate School of Frontier Sciences, The University of Tokyo:Brain Science Institute, RIKEN
Date 2009-01-20
Paper # NC2008-98
Volume (vol) vol.108
Number (no) 383
Page pp.pp.-
#Pages 6
Date of Issue