Explainable AI for Industry 4.0: Semantic Representation of Deep Learning Models

Artificial Intelligence is an important asset of Industry 4.0. Current discoveries within machine learning and particularly in deep enable qualitative change the industrial processes, applications, systems products. However, there challenge related to explainability (and, therefore, trust to) decisions made by models (aka black-boxes) their poor capacity for being integrated with each other. Explainable artificial intelligence needed instead but without loss effectiveness models. In this paper we present transformation technique between black-box explainable (as well as interoperable) classifiers on basis semantic rules via automatic recreation training datasets retraining decision trees (explainable models) between. Our results rule-based good performance efficient process due embedded incremental ignorance discovery adversarial samples (“corner cases”) generation algorithms. We have also shown use-case scenario such interoperable classifiers, which collaborative condition monitoring, diagnostics predictive maintenance distributed (and isolated) smart assets while preserving data knowledge privacy users.