Presentation | 1994/9/22 A Pattern Classification Method Using Domain Knowledge and Empirical Data Kiyoshi Nakabayashi, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | A new classification method is presented which utilizes both human experts′ domain knowledge and empirical data.Based on the Ba yesian learning concept,an equation for estimating a posteriori probability is derived in which the result is given as the weighted-sum of the estimations by the domain knowledge and the training samples.optimization procedures employing likelihood and classification rate as the optimization criteria are provided to determine the confidence factor of the domain knowledge. Experimental results demonstrate that the method achieves better performance than the one only with the domain knowledge or with the training samples. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | Classification / Knowledge Acquisition / A posteriori probability / Bayesian Learning |
Paper # | PRU94-27 |
Date of Issue |
Conference Information | |
Committee | PRU |
---|---|
Conference Date | 1994/9/22(1days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | |
Chair | |
Vice Chair | |
Secretary | |
Assistant |
Paper Information | |
Registration To | Pattern Recognition and Understanding (PRU) |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | A Pattern Classification Method Using Domain Knowledge and Empirical Data |
Sub Title (in English) | |
Keyword(1) | Classification |
Keyword(2) | Knowledge Acquisition |
Keyword(3) | A posteriori probability |
Keyword(4) | Bayesian Learning |
1st Author's Name | Kiyoshi Nakabayashi |
1st Author's Affiliation | NTT Information and Communication Systems Laboratories() |
Date | 1994/9/22 |
Paper # | PRU94-27 |
Volume (vol) | vol.94 |
Number (no) | 242 |
Page | pp.pp.- |
#Pages | 8 |
Date of Issue |