Researchers have developed a new metric called the Explainability-Performance Coefficient (EPC) score to better evaluate the quality of explanations provided by artificial intelligence systems. This model-agnostic metric aims to balance the trade-off between the sparsity of feature selection and the preservation of model performance. Empirical validation across different data types, including tabular, text, and image data, demonstrates that the EPC score effectively identifies operational dependencies within neural networks and aligns with human judgments on explanations. AI
IMPACT This new metric could lead to more reliable and trustworthy AI systems in critical applications by better aligning AI explanations with human understanding.
RANK_REASON The cluster contains a research paper detailing a new metric for evaluating AI explainability. [lever_c_demoted from research: ic=1 ai=1.0]
- alphaXiv
- arXiv
- CatalyzeX
- DagsHub
- EPC score
- Explainability-Performance Coefficient
- explainable AI
- Gotit.pub
- Hugging Face
- IArxiv
- ScienceCast
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
