Lazy Classification Algorithms Based on Deterministic and Inhibitory Rules
Paweł Delimata, Mikhail Moshkov, Andrzej Skowron, Zbigniew Suraj · 2008
In the paper, two families of lazy classification algorithms of polyno-mial time complexity are considered. These algorithms are based on deter-ministic (with a relation “attribute = value ” on the right hand side) and inhibitory (with a relation “at-tribute 6 = value ” on the right hand side) rules, but the direct genera-tion of rules is not required. In-stead of this, the considered algo-rithms extract efficiently for a new object some information about the set of rules which is next used by a decision-making procedure. Results of experiments show that the per-formance of algorithms based on in-hibitory rules is in many cases better than the performance of algorithms based on deterministic rules.