New FrED framework estimates external data influence on AI models

Researchers have developed a new probabilistic framework called FrED to estimate the influence of external data on generative AI models. This black-box method uses a combination of feature similarities and domain-specific knowledge graphs to attribute outputs to specific training data without needing access to model weights. FrED has demonstrated effectiveness in domains like artistic image synthesis and weather forecasting, outperforming standard similarity-based baselines and approaching the performance of gradient-based estimators. AI

IMPACT Provides a more efficient and interpretable method for understanding data attribution in AI models, crucial for transparency and accountability.

RANK_REASON The cluster contains a research paper detailing a new methodology for AI. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• alphaXiv
• arXiv
• CatalyzeX
• DagsHub
• FrED
• Gotit.pub
• Hugging Face
• Iraklis Klampanos
• knowledge graph
• ScienceCast

• paper
• other

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