Reasoning Based on Rules Extracted from Trained Neural Networks via Formal Concept Analysis

Due to their capability of dealing with nonlinear problems, artificial neural networks (ANN) are widely used with several purposes. Once trained, they are also capable of solving unprecedented situations, keeping tolerable errors in their outputs. However, ANN are considered essentially "black boxes". Therefore, humans can not assimilate the knowledge kept by those nets, since such knowledge is implicitly represented by their connection weights. In this paper, a new approach to extract knowledge rules from ANN previously trained through formal concept analysis is presented. The method allows to the knowledge engineer understand the industrial process that is being analyzed, through implications rules of the type if... then. As an example of application a solar energy system is considered. The rules obtained are validated through an expert domain