A spike classification with a Posteriori confidence estimation for ambiguous clusters

Yi Lin Li, Wei Chang Shann, Meng Li Tsai

Research output: Contribution to journalArticlepeer-review

Abstract

We propose a method of classification for two ambiguous clusters of neural action potential signals where the mean and variance of a normally distributed noise are known. The proposed method is based on a fundamental process that classifies a set of two randomly perturbed points. The confidence of the classification is estimated by some computable upper bounds of error rates. We derive the process mathematically, apply it to the task of spike classification, show how it works by a numerical experiment, and compare the result with a PCA scattering plot.

Original languageEnglish
Pages (from-to)133-137
Number of pages5
JournalBiomedical Engineering - Applications, Basis and Communications
Volume20
Issue number2
DOIs
StatePublished - Apr 2008

Keywords

  • Multichannel recording
  • Neural signals
  • Spikes sorting

Fingerprint

Dive into the research topics of 'A spike classification with a Posteriori confidence estimation for ambiguous clusters'. Together they form a unique fingerprint.

Cite this