This paper presents a python library that includes a toolkit with the aim of improving the interpretability of expert systems based on fuzzy cognitive maps through improvements in the visualization and representation of the graphs that can be drawn using the variables of the resulting models. The motivation for the development of the library arises from the need to improve the interpretability of the aforementioned expert systems, given that their multilayer and extracted from experts' knowledge nature can make them very difficult to interpret even for the expert user. Throughout the paper, the reader will be introduced to the basic features of fuzzy logic and fuzzy cognitive maps, and the different developed tools will be defined and exemplified.
The selection of optimal set points is an important problem in modern process control. Fuzzy cognitive maps (FCMs) allow to construct models of complex processes using expert knowledge, which is particularly useful in situations where measuring the variables of interest online is problematic. These models can be used as constraints in optimization problems with the objective of determining optimal set points for those processes. This article presents a reformulation of the constraints imposed by the FCM models that reduces the complexity of the resulting optimization problem and enables the application of heuristic methods for its solution. Computational results show that the use of separable programming on the reformulated problem constitutes a very good alternative, both in terms of solution time and reliability in finding the optimum, enabling the application of FCM modeling to larger systems and easing the practical implementation of the approach.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.