Soft Computing is being widely used in Information Security applications. Particularly, Neuro-Fuzzy approach provides a classification with humanunderstandable rules, yet the accuracy may not be sufficiently high. In this paper we seek for an optimal fuzzy patch configuration that uses elliptic fuzzy patches to automatically extract parameters for the Mamdami-type rules. We proposed a new method based on χ 2 test of data to estimate rotatable patch configuration together with Gaussian membership function. This method has been tested on the automated malware analysis with accuracy up to 92%. Further on, it can find an application in Digital Forensics.