Parallel implementation of evolutionary learning of a fuzzy system with non-singleton fuzzification

Fuzzy systems with fuzzy inputs can be used in tasks where it is necessary to make predictions for data objects that have fizzy characteristics. However, building an optimal block of rules for such a system may be non-trivial, including due to the requirement to have a certain depth of knowledge in the subject area. In this situation, there is a need to automate the process of compiling the rule base, that is, to build a machine learning algorithm. In this paper, we propose to use a genetic (evolutionary) algorithm as such an algorithm. It describes both the specifics of using this family of algorithms for training a fuzzy system, and the features of parallel implementation of the learning process using CUDA technology.