A novel Explainable Artificial Intelligence and secure Artificial Intelligence asset sharing platform for the manufacturing industry
Dimitris Miltiadou, Κonstantinos Perakis, Michele Sesana, Mattia Calabresi, Fenareti Lampathaki, Evmorfia Biliri · 2023
Over the past couple of years, implementations of Artificial Intelligence (AI) have significantly risen in numerous platforms, tools and applications around the world, impacting a broad range of industries such as manufacturing towards Smart Factories and Industry 4.0, in general. Nevertheless, despite industrial AI being the driving force for smart factories, there is strong reluctance in its adoption by manufacturers due to the lack of transparency of the black-box AI models and trust behind the decisions taken, as well as the awareness of where and how it should be incorporated in their processes and products. This paper introduces the Explainable AI platform of XMANAI which takes advantage of the latest AI advancements and technological breakthroughs in Explainable AI (XAI) in order to build "glass box" AI models that are explainable to a "human-in-the-loop" without the decrease of AI performance. The core of the platform consists of a catalogue of hybrid and graph AI models which are built, fine-tuned and validated either as baseline AI models that will be reusable to address any manufacturing problem or trained AI models that have been fine-tuned for solving concrete manufacturing problems in a trustful manner through value-based explanations that are easily and effectively interpreted by humans.