The Empirical Bayes Estimation of an Instantaneous Spike Rate with a Gaussian Process Prior

Posted in Science on August 23, 2008


The Empirical Bayes Estimation of an Instantaneous Spike Rate with a Gaussian Process Prior

Lecture slides:

  • The empirical Bayes estimation of an instantaneous spike rate with a Gaussian process prior
  • Contents
  • Motivation
  • Inference problem
  • Bayesian inference
  • Hyperparameter selection
  • Theoretical analysis: Path integral method
  • Evaluation of the marginal likelihood
  • Result 1: OU process
  • Result 2: Sinusoidal modulation
  • Practical algorithm (using EM)
  • Gaussian approximated EM algorithm
  • Demonstration
  • Summary
  • Acknowledges

Author: Shinsuke Koyama, Kyoto University

Watch Video

Tags: Science, Lectures, Computer Science, Machine Learning, VideoLectures.Net, Gaussian Processes