DEVELOPMENT OF METHODS FOR DECREASING NOISE INFLUENCE ON GENERALIZATION ALGORITHMS

Author(s): Vagin Vadim Nikolaevich, Suvorov Alexander Viktorovich, Fomina Marina Vladimirovna, Morosin Oleg Leonidovich

Rubric: Methodological research

DOI: 10.21777/2312-5500-2016-3-59-68

Release: 2016-3 (15)

Pages: 59-68

Keywords: inductive notion formation; defeasible reasoning; argumentation; justification degrees; non-monotonic reasoning; generalization

Annotation: This paper is devoted to study of the influence of noise in data on the work of generalization algorithms based on building decision trees. Different types of noise and various ways of introducing noise in the learning and test sets are viewed. To improve the efficiency of generalization algorithms, it is proposed to use an argumentation based approach. The results of computer simulation, confirming the effectiveness of the proposed methods and algorithms are presented

Bibliography: Vagin VA.NI., Suvorov AL.VI., Fomina MA.VL., Morosin OL.LE. DEVELOPMENT OF METHODS FOR DECREASING NOISE INFLUENCE ON GENERALIZATION ALGORITHMS // Education Resources and Technologies. – 2016. – № 3 (15). – С. 59-68. doi: 10.21777/2312-5500-2016-3-59-68