Feature Selection Method Based On Hybrid Filter-Metaheuristic Wrapper Approach
High dimension data are often associated with redundant features and there exist many information-theoretic approaches used to select the most relevant set of features and to reduce the feature size. The three most significant approaches are filter, wrap- per, and embedded approaches. Most filter ap...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2020
|
| Subjects: | |
| Online Access: | http://eprints.usm.my/52445/1/Pages%20from%202.%20Final%20Thesis%20Submission.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | High dimension data are often associated with redundant features and there exist many information-theoretic approaches used to select the most relevant set of features and to reduce the feature size. The three most significant approaches are filter, wrap- per, and embedded approaches. Most filter approaches fail to identify the individual contribution of every feature in each set of features in achieving an optimal feature subset. While the wrapper approaches encounter problems from complex interactions among features and stagnation in local optima. To address, these drawbacks, this study investigates filter and wrapper approaches to develop effective hybrid approaches for feature selection. |
|---|