Contributions of machine learning to remote sensing data analysis
This article describes the state of the art on the development and application of machine learning methodologies in the remote sensing domain. In an introductory section, we describe the specific remote sensing analysis problems that are typically handled by machine learning. The remaining of the article is subdivided in a number of sections, dealing with groups of machine learning strategies. The following strategies are elaborated on: kernel methods, neural network methods, manifold learning methods, structured output methods, ensemble learning methods, and sparse learning methods. This specific choice is based on the frequency with which these strategies appeared in the recent remote sensing literature. Each subsection contains a short description of the specific machine learning paradigm and an extensive description of the recent state of the art, with a conceptual description of the new methodologies. We end with some insights in future developments.
| Main Authors: | , , |
|---|---|
| Format: | Part of book or chapter of book biblioteca |
| Language: | English |
| Published: |
Elsevier
|
| Subjects: | Ensemble learning, Hyperspectral image classification, Kernel methods, Machine learning, Manifold learning, Neural networks, Sparse learning, Structured output, |
| Online Access: | https://research.wur.nl/en/publications/contributions-of-machine-learning-to-remote-sensing-data-analysis |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | This article describes the state of the art on the development and application of machine learning methodologies in the remote sensing domain. In an introductory section, we describe the specific remote sensing analysis problems that are typically handled by machine learning. The remaining of the article is subdivided in a number of sections, dealing with groups of machine learning strategies. The following strategies are elaborated on: kernel methods, neural network methods, manifold learning methods, structured output methods, ensemble learning methods, and sparse learning methods. This specific choice is based on the frequency with which these strategies appeared in the recent remote sensing literature. Each subsection contains a short description of the specific machine learning paradigm and an extensive description of the recent state of the art, with a conceptual description of the new methodologies. We end with some insights in future developments. |
|---|