DocumentCode :
923972
Title :
Distribution-free exponential error bound for nearest neighbor pattern classification
Author :
Fritz, Jozsef
Volume :
21
Issue :
5
fYear :
1975
fDate :
9/1/1975 12:00:00 AM
Firstpage :
552
Lastpage :
557
Abstract :
The rate of convergence of the nearest neighbor (NN) rule is investigated when independent identically distributed samples take values in a d -dimensional Euclidean space. The common distribution of the sample points need not be absolutely continuous. An upper bound consisting of two exponential terms is given for the probability of large deviations of error probability from the asymptotic error found by Cover and Hart. The asymptotically dominant first term of this bound is distribution-free, and its negative exponent goes to infinity approximately as fast as the square root of the number of preclassified samples. The second term depends on the underlying distributions, but its exponent is proportional to the sample size. The main term is explicitly given and depends very weakly on the dimension of the space.
Keywords :
Pattern classification; Bayesian methods; Convergence; Error probability; H infinity control; Nearest neighbor searches; Neural networks; Pattern classification; Q measurement; Random variables; Upper bound;
fLanguage :
English
Journal_Title :
Information Theory, IEEE Transactions on
Publisher :
ieee
ISSN :
0018-9448
Type :
jour
DOI :
10.1109/TIT.1975.1055443
Filename :
1055443
Link To Document :
بازگشت