Analogously to the rate distortion function, the error distortion function is defined for a multiple-hypothesis decision problem. The error distortion function e(d) is defined as the supremum of the Bayes’ error probability for transformed observations for which the average distortion is less than d (d > 0) The main result is that the function e(d) is continuous at 0.
ASJC Scopus subject areas
- Information Systems
- Computer Science Applications
- Library and Information Sciences