Reproduce “TimeRSI: Towards Self-Improved Agentic Deployment for Time-Series Foundation Models” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Haoxin Liu, Deqing Fu, Lucheng Fu, Yiqiao Jin, Rushi Qiang, Harshavardhan Kamarthi, B. Aditya Prakash
TimeRSI adapts forecasting pipelines around a frozen time-series foundation model by having a coding agent revise preprocessing, features, and post-processing using historical forecast errors, then select among candidates using completed similar windows. In a parking task, MASE fell from 0.678730 to 0.641658. Across benchmark comparisons it ranked first on six metrics, but future distribution shift remains unresolved.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.