Semi-supervised Classification with Modified Kernel Partial Least Squares
| Autor: | Paweł Błaszczyk |
|---|---|
| Rok vydání: | 2017 |
| Předmět: |
Computer science
business.industry Pattern recognition Semi-supervised learning Kernel principal component analysis Support vector machine Iteratively reweighted least squares ComputingMethodologies_PATTERNRECOGNITION Kernel method Kernel (statistics) Least squares support vector machine Principal component regression Artificial intelligence business |
| Zdroj: | Transactions on Engineering Technologies ISBN: 9789811027161 |
| Popis: | The aim of this paper is to present a new semi-supervised classification method based on modified Partial Least Squares algorithm and Gaussian Mixture Models. Combining the information contained in unlabeled samples together with the available training labeled samples can increase the classification performance. Our method relies on combining two kernel functions: the standard kernel calculated on data from labeled samples and a generative kernel directly learned by clustering the data. The economical datasets are used to compare the performance of the classification. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|