Name
Abstract | ||
|---|---|---|
Abstract In many rule induction algorithms, it is relatively common that a large number of rules is generated This complexity may harm the comprehensibility of the model without improving its predictive performance For aiding automatic knowledge acquisition and knowledge discovery in classification domains, we propose a framework to post - process Boolean rules obtained from rule induction algorithms This process generates probabilistic rule sets with fewer rules and premises, while maintaining comparable classification accuracy |
Year | Venue | Keywords |
|---|---|---|
2000 | ICML | hybrid learning,simplified rule bases,rule based |
Field | DocType | ISBN |
Pattern recognition,Computer science,Rule induction,Artificial intelligence,Artificial neural network,Machine learning | Conference | 1-55860-707-2 |
Citations | PageRank | References |
1 | 0.35 | 10 |
Authors | ||
2 | ||
Name | Order | Citations | PageRank |
|---|---|---|---|
| Ricardo Bezerra de Andrade e Silva | 1 | 109 | 24.56 |
| Teresa Bernarda Ludermir | 2 | 928 | 108.14 |