New analysis reveals accuracy of AI data attribution methods

Researchers have developed a new mathematical analysis for data attribution methods like Influence Functions (IF) and Newton Step (NS) in convex learning problems. This analysis does not rely on strong convexity assumptions and provides tighter bounds, addressing limitations of previous work. The findings explain why NS data attribution is often more accurate than IF and establish asymptotic scaling laws for their errors. AI

影响 Provides a more accurate understanding of data attribution methods, crucial for model interpretability and responsible AI development.

排序理由 Academic paper detailing a new analysis of existing machine learning methods. [lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv stat.ML 阅读 →

• Influence Functions
• Ittai Rubinstein
• Newton Step

• 论文
• 其他

AI 生成摘要 · Google Gemini · 来自 1 个来源。 我们如何撰写摘要 →