arXiv:2606.24974v1 Announce Type: new Abstract: Explainability techniques are used to assess the output of various deep learning models. This is especially true in healthcare, where models need to be trusted and decisions justified. Explainability (XAI) tools use heuristics which often add signal noise to the explanation "core". It is not always obvious what is signal from the model and what is noise from the XAI. We propose the use of spectral entropy as a measure of noise in XAI output. We demonstrate its usefulness in the context of classifying arrhythmias in an ECG dataset with different p
The proliferation of deep learning models in critical domains like healthcare necessitates robust explainability, leading to increased research into XAI method reliability.
Quantifying XAI-introduced noise is crucial for building trustworthy AI systems, particularly in sensitive applications where erroneous explanations could have severe consequences.
This research provides a concrete method to assess the quality of XAI outputs, enabling better selection and refinement of explainability techniques for real-world deployments.
- · AI developers
- · Healthcare sector
- · Regulatory bodies
- · Patients
- · Untrustworthy XAI methods
- · Blind AI adoption
Improved reliability and interpretability of AI models in healthcare.
Accelerated adoption of AI in regulated industries due to increased trust and verifiable explanations.
Potential for new standards and metrics around AI explainability and trustworthiness, transforming AI auditing practices.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.LG