New DEFT framework enhances expert guidance for time-series models

Researchers have developed DEFT, a new framework for expert-guided forecast editing in time-series foundation models. This method addresses the limitation that existing models produce fixed forecasts and cannot incorporate expert feedback directly. DEFT balances exploiting the model's predictive distribution with exploring new forecast trajectories by refining trend and seasonal components, thereby making more efficient use of limited expert queries. AI

IMPACT This framework could improve the adaptability and accuracy of time-series forecasting models by enabling more effective integration of domain-specific knowledge.

RANK_REASON The cluster contains a research paper detailing a new framework for time-series foundation models. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• arXiv
• Bayesian optimization
• cross-entropy methods
• Hugging Face

• paper
• model release

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