Linear discriminant analysis (LDA) is a basic tool of pattern recognition, and it is used in extensive fields, e.g. face identification. However, LDA is poor at adaptability since it is a batch type algorithm. To overcome this, new algorithms of online LDA are proposed in the present paper. In face identification task, it is experimentally shown that the new algorithms are about two times faster than the previously proposed algorithm in terms of the number of required examples, while the previous algorithm attains better final performance than the new algorithms after sufficient steps of learning. The meaning of new algorithms are also discussed theoretically, and they are suggested to be corresponding to combination of PCA and Mahalanobis distance.
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copy
Kazuyuki HIRAOKA, Masashi HAMAHIRA, Ken-ichi HIDAI, Hiroshi MIZOGUCHI, Taketoshi MISHIMA, Shuji YOSHIZAWA, "Fast Algorithm for Online Linear Discriminant Analysis" in IEICE TRANSACTIONS on Fundamentals,
vol. E84-A, no. 6, pp. 1431-1441, June 2001, doi: .
Abstract: Linear discriminant analysis (LDA) is a basic tool of pattern recognition, and it is used in extensive fields, e.g. face identification. However, LDA is poor at adaptability since it is a batch type algorithm. To overcome this, new algorithms of online LDA are proposed in the present paper. In face identification task, it is experimentally shown that the new algorithms are about two times faster than the previously proposed algorithm in terms of the number of required examples, while the previous algorithm attains better final performance than the new algorithms after sufficient steps of learning. The meaning of new algorithms are also discussed theoretically, and they are suggested to be corresponding to combination of PCA and Mahalanobis distance.
URL: https://global.ieice.org/en_transactions/fundamentals/10.1587/e84-a_6_1431/_p
Copy
@ARTICLE{e84-a_6_1431,
author={Kazuyuki HIRAOKA, Masashi HAMAHIRA, Ken-ichi HIDAI, Hiroshi MIZOGUCHI, Taketoshi MISHIMA, Shuji YOSHIZAWA, },
journal={IEICE TRANSACTIONS on Fundamentals},
title={Fast Algorithm for Online Linear Discriminant Analysis},
year={2001},
volume={E84-A},
number={6},
pages={1431-1441},
abstract={Linear discriminant analysis (LDA) is a basic tool of pattern recognition, and it is used in extensive fields, e.g. face identification. However, LDA is poor at adaptability since it is a batch type algorithm. To overcome this, new algorithms of online LDA are proposed in the present paper. In face identification task, it is experimentally shown that the new algorithms are about two times faster than the previously proposed algorithm in terms of the number of required examples, while the previous algorithm attains better final performance than the new algorithms after sufficient steps of learning. The meaning of new algorithms are also discussed theoretically, and they are suggested to be corresponding to combination of PCA and Mahalanobis distance.},
keywords={},
doi={},
ISSN={},
month={June},}
Copy
TY - JOUR
TI - Fast Algorithm for Online Linear Discriminant Analysis
T2 - IEICE TRANSACTIONS on Fundamentals
SP - 1431
EP - 1441
AU - Kazuyuki HIRAOKA
AU - Masashi HAMAHIRA
AU - Ken-ichi HIDAI
AU - Hiroshi MIZOGUCHI
AU - Taketoshi MISHIMA
AU - Shuji YOSHIZAWA
PY - 2001
DO -
JO - IEICE TRANSACTIONS on Fundamentals
SN -
VL - E84-A
IS - 6
JA - IEICE TRANSACTIONS on Fundamentals
Y1 - June 2001
AB - Linear discriminant analysis (LDA) is a basic tool of pattern recognition, and it is used in extensive fields, e.g. face identification. However, LDA is poor at adaptability since it is a batch type algorithm. To overcome this, new algorithms of online LDA are proposed in the present paper. In face identification task, it is experimentally shown that the new algorithms are about two times faster than the previously proposed algorithm in terms of the number of required examples, while the previous algorithm attains better final performance than the new algorithms after sufficient steps of learning. The meaning of new algorithms are also discussed theoretically, and they are suggested to be corresponding to combination of PCA and Mahalanobis distance.
ER -