Fuzzy probabilistic rule induction
G.C. van den Eijkel, Eric Backer · 2003
The paper presents a novel uncertainty framework for rule induction from examples: a fuzzy probabilistic framework. The main motivation for this framework stems from the problem of data fit vs. mental fit in knowledge acquisition for decision support systems. The framework is based on an extension of the probability of a fuzzy event (as defined by L.A. Zadeh, 1968)), and is highly suitable for learning and reasoning with uncertainty. Experiments show that the framework results in a simple rule base by which highly accurate classifications are obtained and explained.