Democratizing AI for Condition-Based Maintenance Leveraging Probabilistic Programming for Symbolic Reasoning

SUMMARY AND CONCLUSIONSAdvances in AI/ML have demonstrated enormous potential in improving and optimizing condition-based maintenance processes; however, AI/ML solutions themselves inevitably become a maintenance liability, wherein the end users must repeatedly work with a data scientist to update the AI/ML application. In this paper, we present novel research that alleviates this problem by enabling non-AI/ML experts to create and maintain AI/ML applications with minimal guidance from a data scientist. We describe this research as the "Democratization" of AI/ML and accomplish this by leveraging cutting edge techniques in knowledge capture and probabilistic programming as a way to represent higher level symbolic reasoning and delegating sub-symbolic reasoning to traditional data driven approaches.