Using SAT/SMT Solvers for Efficiently Tuning Fuzzy Logic Programs

During the last years we have developed advanced tools for tuning fuzzy logic programs devoted to facilitate the selection of the more appropriate set of weights and fuzzy connectives used in programs rules. Designing accurate techniques for automating these tasks is very useful for programmers, even when they are time consuming. In order to increase its performance, in this paper we make use of powerful and well-known SAT/SMT solvers for improving our original approaches. Inspired by some previous experiences we have acquired in this setting, whose impact is growing in many modern software tools, we show some representative experiments (related to circuit validation and linear regression) and benchmarks which illustrate the significant advantages enjoyed by the new empowered method.