On the minimal dimension of sufficient statistics

  • Authors:

  • L. J. Buturovic

  • Affiliations:

  • Inst. of Appl. Math. & Electron., Belgrade

  • Venue:
  • IEEE Transactions on Information Theory
  • Year:
  • 2006

Quantified Score

Hi-index 754.84

Visualization

Abstract

The minimal dimension of sufficient continuous statistics for a finite family of continuous probability density functions is determined. The relation between sufficient statistics and optimal statistics in the context of pattern recognition systems is studied, and the application of the main result to the case of exponential families is discussed