Explainable, Physics Aware, Trustworthy AI Paradigm Shift for Synthetic Aperture Radar

Physics-informed AI improves trustworthiness: The paper proposes grounding AI explainability in established physical models of how SAR systems work—including imaging geometry, signal formation, and scattering properties—rather than treating AI as a "black box." This hybrid approach of combining physics with machine learning is intended to make AI decisions more interpretable and trustworthy for users.

Domain expertise should guide AI training: The authors argue that explainable data transformations based on well-established SAR physics should be used to generate training inputs and provide "knowledgeable feedback" during model development. This means domain experts should actively shape how AI systems learn, rather than letting algorithms discover patterns independently.

Transparency requires multiple layers of explanation: The paper identifies three physical layers (instrument/platform, imaging formation, and scattering signatures) that should be integrated into AI models. By decomposing complex systems into these understandable components, the approach aims to provide "supplementary explanations" that help stakeholders understand why the AI reached particular conclusions.

This framework extends beyond radar: The authors note their proposed concept has "applicability to the entire class of coherent imaging sensors and other computational imaging systems," suggesting that physics-aware, explainable AI could become a broader standard for trustworthy AI in technical imaging applications.