New CHARTER framework improves evaluation in computational pathology

A new evaluation framework called CHARTER has been introduced to address issues in computational pathology where the reference used for evaluating predictions can inadvertently alter the results. CHARTER aims to make these dependencies explicit by requiring researchers to declare their intended target and reference, quantify shifts in predictions caused by candidate filtering, and audit the stability of comparative conclusions. This framework helps differentiate genuine preservation of predictions from apparent gains that arise from changing the reference being explained, as demonstrated by significant reversals observed in audits. AI

IMPACT This framework could improve the reliability and comparability of AI model evaluations in computational pathology.

RANK_REASON The item is an academic paper detailing a new evaluation framework for a specific technical domain. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• AC Milan
• alphaXiv
• arXiv
• CatalyzeX Code Finder for Papers
• CHARTER
• Computational Pathology
• Connected Papers
• CORE Recommender
• DagsHub
• Gotit.pub
• Hugging Face
• Influence Flower
• Litmaps
• Random Killing
• ScienceCast
• scite Smart Citations

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →