Two-way concept-cognitive learning method: a perspective from progressive learning of fuzzy skills

Abstract

ABSTRACT Concept-cognitive learning and two-way concept-cognitive learning have been widely used to learn concepts from given cues, and fruitful results have been obtained. However, there are still some limitations: (i) The existing two-way concept-cognitive learning methods based on competences focused on dichotomous skills, but they did not consider the proficiency of skills in the concept-cognitive learning process and ignored the dynamic learning of skills. (ii) The existing necessary and sufficient information granules based on competences cannot be learned directly from any information granule. (iii) Most of the other existing two-way concept-cognitive learning models in fuzzy formal context have been studied through an antitone Galois connection, but they cannot directly deal with an isotone Galois connection between items and fuzzy skills. To overcome these issues, we propose a novel two-way concept-cognitive learning model based on fuzzy skills to learn fuzzy skill-based cognitive concepts (i.e. necessary and sufficient fuzzy skill-based information granules) from a perspective of skill proficiency. And we optimize the two-way concept-cognitive learning model to learn directly fuzzy skill-based cognitive concepts from any information granule and save time for learning concepts. Furthermore, a progressive learning mechanism is explored to deal with dynamic data. The experimental results of 12 UCI data sets show that the proposed concept-cognitive learning models in this paper are feasible and effective.