Readable Twins of Unreadable Models
This addresses the need for explainable AI systems, particularly in deep learning, though it appears incremental by adapting digital twin concepts to a new domain.
The paper tackles the problem of explainability in deep learning by proposing readable twins as imprecise information flow models for unreadable models, with an example applied to MNIST digit classification.
Creating responsible artificial intelligence (AI) systems is an important issue in contemporary research and development of works on AI. One of the characteristics of responsible AI systems is their explainability. In the paper, we are interested in explainable deep learning (XDL) systems. On the basis of the creation of digital twins of physical objects, we introduce the idea of creating readable twins (in the form of imprecise information flow models) for unreadable deep learning models. The complete procedure for switching from the deep learning model (DLM) to the imprecise information flow model (IIFM) is presented. The proposed approach is illustrated with an example of a deep learning classification model for image recognition of handwritten digits from the MNIST data set.