Title :
On optimal sequential prediction for general processes
Author :
Nobel, Andrew B.
Author_Institution :
Dept. of Stat., Univ. of North Carolina, Chapel Hill, NC, USA
fDate :
1/1/2003 12:00:00 AM
Abstract :
This paper considers several aspects of the sequential prediction problem for unbounded, nonstationary processes under pth power loss ℓ
p(u,v)=|u-v|
p, 1
\n\n\t\t
Keywords :
binary sequences; calibration; optimisation; prediction theory; Bayes prediction; Cesaro optimal prediction schemes; Hamming loss; additive noise; asymptotic performance; binary processes; binary sequences; calibration; ergodic processes; existence results; general processes; generalized prediction; nonstationary process; optimal binary prediction; optimal sequential prediction; power loss; sequential prediction problem; squared loss; thresholding; unbounded process; uniqueness results; weak calibration; Additive noise; Binary sequences; Calibration; H infinity control; Helium; Neural networks; Random variables; Statistics; Stochastic processes;
Journal_Title :
Information Theory, IEEE Transactions on
DOI :
10.1109/TIT.2002.806141
