Smace Preprint

New preprint available: we propose a new method for the explainability of composite AI systems