Good-turing estimation for the frequentist n-tuple classifier

Abstract

We present results concerning the application of the Good-Turing (GT) estimation method to the frequentist n-tuple system. We show that the Good-Turing method can, to a certain extent rectify the Zero Frequency Problem by providing, within a formal framework, improved estimates of small tallies. We also show that it leads to better tuple system performance than Maximum Likelihood estimation (MLE). However, preliminary experimental results suggest that replacing zero tallies with an arbitrary constant close to zero before MLE yields better performance than that of GT system.

Divisions: Aston University (General)
Event Title: Proceedings of the Weightless Neural Network Workshop 1995, Computing with Logical Neurons
Event Type: Other
Event Dates: 1995-09-01 - 1995-09-01
Uncontrolled Keywords: good-turing,zero frequency,estimates,maximum likelihood estimation
Last Modified: 12 Jan 2024 08:12
Date Deposited: 15 Jul 2009 08:39
PURE Output Type: Chapter
Published Date: 1995-09
Authors: Morciniec, Michal
Rohwer, Richard

Download

Export / Share Citation


Statistics

Additional statistics for this record