Explainable AI Through a Democratic Lens: DhondtXAI for D'Hondt-Projected Feature Attribution
arXiv:2411.05196v3 Announce Type: replace-cross Abstract: This study presents DhondtXAI as a SHAP-independent, D'Hondt-based attribution framework for tabular XAI. Instead of model-native feature importance or SHAP values, DhondtXAI computes background-interventional removal effects, separates positive and negative evidence, forms optional feature alliances, applies optional thresholds, allocates seats via the D'Hondt rule, and projects onto the local model-output difference. Completeness is preserved by construction, with the projection residual ratio reported as a diagnostic. The method is e
The increasing complexity and opacity of AI models necessitate more robust and interpretable explainability frameworks, particularly as AI adoption grows in critical applications.
Improved explainable AI (XAI) methods are crucial for building trust, ensuring accountability, and enabling debugging and regulatory compliance in AI systems.
This new D'Hondt-based attribution framework offers an alternative to existing XAI methods like SHAP, providing a potentially more robust and transparent way to understand feature contributions to AI decisions.
- · AI ethicists
- · Regulatory bodies
- · Developers of high-stakes AI applications
- · Researchers in explainable AI
- · Opaque AI models
- · Companies unable to explain their AI systems
DhondtXAI offers a novel, SHAP-independent method for feature attribution in tabular explainable AI.
Broader adoption of such methods could lead to higher standards for AI explainability and a demand for AI systems that inherently support transparency.
This could accelerate the creation of more trustworthy and auditable AI, potentially influencing public policy and regulatory frameworks globally.
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