Rule induction with CN2: some recent improvements
EWSL-91 Proceedings of the European working session on learning on Machine learning
Advances in knowledge discovery and data mining
Advances in knowledge discovery and data mining
INLEN: a methodology and integrated system for knowledge discovery in databases
INLEN: a methodology and integrated system for knowledge discovery in databases
Machine Learning
Machine Learning
A Pathology of Bottom-Up Hill-Climbing in Inductive Rule Learning
ALT '02 Proceedings of the 13th International Conference on Algorithmic Learning Theory
Traffic Prediction for Agent Route Planning
ICCS '08 Proceedings of the 8th international conference on Computational Science, Part III
Measures of Ruleset Quality Capable to Represent Uncertain Validity
ECSQARU '07 Proceedings of the 9th European Conference on Symbolic and Quantitative Approaches to Reasoning with Uncertainty
Study on the Non-expandability of DNF and Its Application to Incremental Induction
ICIRA '08 Proceedings of the First International Conference on Intelligent Robotics and Applications: Part I
Measures of ruleset quality for general rules extraction methods
International Journal of Approximate Reasoning
Conflict-free incremental learning
ICIC'09 Proceedings of the Intelligent computing 5th international conference on Emerging intelligent computing technology and applications
Hi-index | 0.00 |
In concept learning and data mining tasks, the learner istypically faced with a choice of many possible hypotheses or patternscharacterizing the input data. If one can assume that training datacontain no noise, then the primary conditions a hypothesis mustsatisfy are consistency and completeness with regard to the data. Inreal-world applications, however, data are often noisy, and theinsistence on the full completeness and consistency of the hypothesisis no longer valid. In such situations, the problem is to determine ahypothesis that represents the best trade-off between completenessand consistency. This paper presents an approach to this problem inwhich a learner seeks rules optimizing a rule qualitycriterion that combines the rule coverage (a measure ofcompleteness) and training accuracy (a measure of inconsistency).These factors are combined into a single rule quality measure througha lexicographical evaluation functional (LEF). The method hasbeen implemented in the AQ18 learning system for natural inductionand pattern discovery, and compared with several other methods.Experiments have shown that the proposed method can be easilytailored to different problems and can simulate different rulelearners by modifying the parameter of the rule quality criterion.