Diversity-based case base maintenance
- Resource Type
- Conference
- Authors
- Lan-Zhen Yang; Ming-Hu Ha; Xi-Zhao Wang
- Source
- Proceedings of the 2003 International Conference on Machine Learning and Cybernetics (IEEE Cat. No.03EX693) Machine learning and cybernetics Machine Learning and Cybernetics, 2003 International Conference on. 3:1591-1596 Vol.3 2003
- Subject
- Computing and Processing
Robotics and Control Systems
Computer aided software engineering
Machine learning algorithms
Size control
Mathematics
Computer science
Electronic switching systems
System performance
Problem-solving
Control systems
Equations
- Language
Three algorithms are proposed for maintaining a case base. Different from other algorithms that have been proposed for updating a case base, the diversity of cases (relative to each other) plays an important role in these new algorithms. Specifically, these algorithms combine similarity and diversity of cases together for determining a new case base. Experimental results demonstrate that these algorithms can greatly reduce the size of a case base without comprising its competence, especially for the algorithm based on evolution strategies (ESs).