Theoretical and practical considerations of uncertainty and complexity in automated knowledge acquisition

摘要: Inductive machine learning has become an important approach to automated knowledge acquisition from databases. The disjunctive normal form (DNF), as the common analytic representation of decision trees and tables (rules), provides a basis for formal analysis uncertainty complexity in inductive learning. A theory general is developed based on C. Shannon's (1949) expansion discrete DNF, probabilistic induction system PIK further extracting real world data. Then we combine practical approaches study how data characteristics affect Three characteristics, namely, disjunctiveness, noise incompleteness, are studied. combination leveled pruning, condensing resampling estimation turns out be very powerful method dealing with highly inadequate Finally compared other recent systems number domains. >