| Presentation | 2001/10/11 Discriminative Learning to Improve the Outlier Resistance of Neural Classifiers Cheng-Lin Liu, Hiroshi Sako, |
|---|---|
| PDF Download Page | PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | The resistance to outliers is important for pattern field recognition integrating segmentation and classification. In this paper, we propose a discriminative learning method for training neural classifiers with outliers so as to improve the outlier resistance. The proposed method is an extension of the minimum classification error(MCE) method of Katagiri and Juang. To learn with both object patterns and outlier patterns, the MCE criterion is modified to consider not only the classification error, but also the false rejection of object patterns and the false acceptance of outliers. We have applied this method to four neural classifiers, namely, single-layer perceptron(SLP), multi-layer perceptron(MLP), polynomial classifier(PC), and radial basis function(RBF) classifier. Experiments were conducted on handwritten digit recognition with synthesized outliers and non-character images collected in character string segmentation. The results show that in respect of the tradeoff between the reject of object patterns and the false acceptance of outliers, the performance of the proposed method is superior to that of MSE(minimum square error) learning. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Pattern classification / neural networks / outlier rejection / discriminative learning with outliers / handwritten digit recognition |
| Paper # | PRMU2001-99,NC2001-49 |
| Date of Issue |
| Conference Information | |
| Committee | PRMU |
|---|---|
| Conference Date | 2001/10/11(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 Media Understanding (PRMU) |
|---|---|
| Language | ENG |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Discriminative Learning to Improve the Outlier Resistance of Neural Classifiers |
| Sub Title (in English) | |
| Keyword(1) | Pattern classification |
| Keyword(2) | neural networks |
| Keyword(3) | outlier rejection |
| Keyword(4) | discriminative learning with outliers |
| Keyword(5) | handwritten digit recognition |
| 1st Author's Name | Cheng-Lin Liu |
| 1st Author's Affiliation | Central Research Laboratory, Hitachi, Ltd.() |
| 2nd Author's Name | Hiroshi Sako |
| 2nd Author's Affiliation | Central Research Laboratory, Hitachi, Ltd. |
| Date | 2001/10/11 |
| Paper # | PRMU2001-99,NC2001-49 |
| Volume (vol) | vol.101 |
| Number (no) | 362 |
| Page | pp.pp.- |
| #Pages | 8 |
| Date of Issue | |