| C04-1190 | al. , 1998 ) . Thus we train the | KPCA | model using the following algorithm |
| C04-1190 | class prediction . What permits | KPCA | to apply stronger generalization |
| C04-1190 | features can be identified during | KPCA | training . 4.2 Algorithm The |
| C04-1190 | in general , and not just the | KPCA | - based model . Put another way |
| C04-1190 | of our WSD model is that during | KPCA | training , the sense class is |
| C04-1190 | to generalize . The nature of | KPCA | , however , suggests a strategy |
| C04-1190 | KPCA model from the supervised | KPCA | baseline model described above |
| C04-1190 | difference of the semi-supervised | KPCA | model from the supervised KPCA |
| C04-1190 | unannotated data , with which the | KPCA | model can first be trained in |
| C04-1190 | nonlinear mapping ; in this respect | KPCA | is similar to Support Vector |
| C04-1190 | Ph ( xj ) ) . In the supervised | KPCA | model , training vectors such |
| C04-1190 | Component Analysis technique , or | KPCA | , is a method of nonlinear principal |
| C04-1190 | Put another way , even though | KPCA | is able to generalize over combinations |
| C04-1190 | corresponding vectors using the trained | KPCA | model and classify the resultant |
| C04-1190 | lth element of ^ al . 3.2 Why is | KPCA | suited to WSD ? The potential |
| C04-1190 | special case , which may explain why | KPCA | always outperforms PCA . 4 Semi-supervised |
| C04-1190 | Principal Component Analysis ( | KPCA | ) technique to make predictions |
| C04-1190 | dimensional vector spaces . Since the | KPCA | transform is computed from unsupervised |
| C04-1190 | components as transformed via PCA and | KPCA | . Observed vectors PCA-transformed |
| C04-1190 | nonlinear transform obtained via | KPCA | as described below . 2001 ) , |
