Get Started
Home
Topics
Search
Library
Research questionHow can online time-series forecasters adapt to evolving, unseen patterns without costly continual parameter updates?Online forecasters must make predictions as data distributions change, including when new patterns emerge. Repeatedly updating many model parameters can be costly and may not support sustained adaptation.
AI
Machine Learning
Research Paper
Statistical Machine Learning
Technology
Latest papersRecent research connected to this question, newest first.Compositional Spectral Prompts for LLM-based Online Time Series ForecastingThe source studies an LLM-based forecaster with a frozen backbone and compositional spectral prompts derived from frequency-domain bases. Evidence comes from experiments on real-world datasets involving extended online phases and cross-dataset distribution shifts.research paper · Sep 2, 2026
Related questions
How can lightweight time-series forecasters learn accurate, specialized forecasts from scarce private data?How can recommenders adapt to changing preferences without trusting unreliable inputs?How can parametric PDE solution operators extrapolate to unseen regimes without failing silently?How can zero-shot building IoT forecasting adapt covariates to each building’s physical topology?