DocumentCode :
1174522
Title :
Asymptotically optimal classification for multiple tests with empirically observed statistics
Author :
Gutman, Michael
Author_Institution :
Technion-Israel Inst. of Technol., Haifa, Israel
Volume :
35
Issue :
2
fYear :
1989
fDate :
3/1/1989 12:00:00 AM
Firstpage :
401
Lastpage :
408
Abstract :
The decision problem of testing M hypotheses when the source is Kth-order Markov and there are M (or fewer) training sequences of length N and a single test sequence of length n is considered. K, M, n, N are all given. It is shown what the requirements are on M , n, N to achieve vanishing (exponential) error probabilities and how to determine or bound the exponent. A likelihood ratio test that is allowed to produce a no-match decision is shown to provide asymptotically optimal error probabilities and minimum no-match decisions. As an important serial case, the binary hypotheses problem without rejection is discussed. It is shown that, for this configuration, only one training sequence is needed to achieve an asymptotically optimal test
Keywords :
error statistics; information theory; probability; statistical analysis; Kth-order Markov source; M-hypotheses problem; asymptotically optimal classification; binary hypotheses problem; decision problem; empirically observed statistics; error probabilities; information theory; likelihood ratio test; multiple tests; no-match decision; training sequences; Computer errors; Data compression; Error probability; Information theory; Statistical analysis; Testing;
fLanguage :
English
Journal_Title :
Information Theory, IEEE Transactions on
Publisher :
ieee
ISSN :
0018-9448
Type :
jour
DOI :
10.1109/18.32134
Filename :
32134
Link To Document :
بازگشت