EXPLAIN

To realize the possible benefits of AI in industrial applications, domain experts and end-users need to be given insights into the internal reasoning of ML-models. This is the subject of XAI research which gained a lot of attention in the last years. However, the needs and characteristics of the users and their working context is very diverse in industrial applications and these aspects need to be considered when designing industrial AI systems. Hence, it is imperative to consider diverse industrial users, use-cases, and data to develop a better understanding of the context and requirement of appropriate AI solutions. This necessitates close collaboration between AI use-case providers, ai-providers, and research actors from machine learning, XAI, software engineering, user experience and human factors. The proposed project EXPLAIN seeks to realize an end-to-end ML lifecycle that is interactive and explainable for industrial domain experts. This entails involving domain experts deeply by introducing four new steps in the lifecycle. In an (1) Explanatory Training phase, together with ML experts, the domain experts will directly interact with the ML-model during the training process and receive explanations of the model output and be able to give feedback. In an (2) Explanation Review, the validation of ML solutions is done by enabling domain experts to gain insights into the internal reasoning of the trained model to ensure that relevant concepts are learned from the provided data. For the end-users, every (3) output of the ML-models is explained and end-users can also (4) give feedback and trigger an incremental explanatory training.