DEFT Improves Expert-Guided Time-Series Forecast Editing

Hung Le, Minh Hoang Nguyen, Manh Nguyen, Huu Hiep Nguyen, Dai Do· July 23, 2026 View original

Summary

DEFT is a new framework that enhances expert-guided forecast editing for time-series foundation models by balancing exploitation of the model's predictive distribution with structured exploration in a decomposed trend-seasonal space.

This research introduces DEFT, a novel framework for expert-guided forecast editing specifically designed for time-series foundation models. While these models can generate forecasts across diverse domains without task-specific training, their outputs are typically fixed and cannot directly incorporate expert feedback. DEFT addresses this by allowing an expensive expert evaluator to score candidate future trajectories, guiding forecast revisions under a tight query budget. DEFT strategically balances two extreme approaches: purely exploiting the foundation model's predictive distribution (best-of-N) and broadly exploring the forecast horizon as an unstructured vector. It first exploits the model's samples in a decomposed trend-seasonal space, then refines these components through structured exploration. The framework reuses expert scores for individual trend and seasonal components that appear in queried recombinations, maximizing the value of each expert query. Across 78 datasets, three foundation models, and various feedback types, DEFT consistently improved the effectiveness of expert guidance, with a molecular dynamics case study suggesting its applicability to physically grounded feedback.

Why it matters

Professionals relying on time-series forecasts, especially in critical domains, can use DEFT to incorporate valuable expert insights more efficiently, leading to more accurate and trustworthy predictions from foundation models.

How to implement this in your domain

  1. 1Evaluate DEFT's approach for integrating expert feedback into your time-series forecasting workflows.
  2. 2Decompose your time-series forecasts into trend and seasonal components for more structured expert interaction.
  3. 3Design expert feedback mechanisms that can score complete trajectories while providing component-level insights.
  4. 4Compare DEFT's performance against simpler best-of-N or direct optimization methods under limited expert query budgets.

Who benefits

FinanceSupply ChainEnergyHealthcareManufacturing

Key takeaways

  • DEFT enables efficient expert-guided editing of time-series foundation model forecasts.
  • It balances exploiting model predictions with structured exploration in a decomposed space.
  • Expert queries are maximized by reusing scores for trend and seasonal components.
  • DEFT consistently outperforms direct search methods across various benchmarks.

Original post by Hung Le, Minh Hoang Nguyen, Manh Nguyen, Huu Hiep Nguyen, Dai Do

"arXiv:2607.19659v1 Announce Type: new Abstract: Time-series foundation models can forecast across heterogeneous domains without task-specific training, but their forecasts are fixed once produced and cannot directly incorporate task-specific expert feedback. We study expert-guide…"

View on X

Originally posted by Hung Le, Minh Hoang Nguyen, Manh Nguyen, Huu Hiep Nguyen, Dai Do on X · view source

Want to go deeper?

Turn these trends into skills with Learnijoy's hands-on AI & tech courses.

Explore courses