Created on November 16, 2021
2021
New preprint available: we propose a new method for the explainability of composite AI systems