Talking to Blackbox: Explainability Through P+NP=1

Abstract excerpt

The interpretability of complex AI systems remains one of the most critical challenges for modern machine learning, particularly when dealing with blackbox models such as deep neural networks and large language models (LLMs). While current Explainable AI (XAI) techniques — notably SHAP and LIME — provide local or feature-based insights, they often rely on additive approximations that fail to capture the underlying...

Topics

Open a Topic to create a Post that cites this publication.

• Artificial Intelligence
• Decision Support Systems, Clinical
• Delivery of Health Care
• Electronic Health Records
• Forecasting
• Genetic Predisposition to Disease
• Genetic Variation
• Genetics, Medical

Identifiers and source

Literature Corpus work

3e89c290-4117-5964-9683-44691c6fda3e

DOI

10.20944/preprints202507.1759.v1