Learning with relevant features and examples

In this paper we focus on selection of relevant features and examples, which is one of the central problems in machine learning and pattern recognition. We describe a way of selecting all combinations of relevant, irredundant features of training examples, and possible ways to identify a relevant, irredundant features combination of the target concept. We also propose a new example selection method which is based on the filtering of the so called pattern frequency domain and which resembles frequency domain filtering in signal and image processing. The empirical results show the effectiveness of the proposed selection methods for relevant features and examples.