Title :
Identification of sample size needed in a given classification task
Author_Institution :
Dept. of Electr. Eng., Ngee Ann Polytech., Singapore
Abstract :
The determination of a sufficient size for classification tasks has been haphazard and lacking in formal mathematical analysis. Instead of using current PAC learning, this paper discusses another approach to tackling this problem. It involves 3 steps: estimating the sample means, determining if the sample means are significantly different from each other and determining if the standard error of the estimated proportion is sufficiently small
Keywords :
pattern classification; statistical analysis; classification task; sample means; sample size; standard error; Equations; Error analysis; Error correction; Estimation error; Intelligent networks; Mathematical analysis; Neural networks; State estimation; System testing;
Conference_Titel :
Neural Networks, 1995. Proceedings., IEEE International Conference on
Conference_Location :
Perth, WA
Print_ISBN :
0-7803-2768-3
DOI :
10.1109/ICNN.1995.487875
