Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
6905205 | Applied Soft Computing | 2015 | 14 Pages |
Abstract
- A new operational framework to compromise between interpretability and accuracy is proposed.
- The proposed Evolving Information Granule (EIG) increases the information granules (rules) by partitioning the output domain.
- The distinct output-contexts are adapted using a constraint which has the dynamic nature.
- The overfitting criterion is estimated online and approximated from the current and previous evolving stage.
- The results show that the proposed model achieves a reasonable accuracy with lower number of rules.
Keywords
Related Topics
Physical Sciences and Engineering
Computer Science
Computer Science Applications
Authors
Md. Manjur Ahmed, Nor Ashidi Mat Isa,