Compositional Spectral Prompts for LLM-based Online Time Series Forecasting
cs.LG
Submitted: 2026-09-02
Updated: 2026-09-02
Comments: CIKM 2026
Code: https://github.com/seungyoon-Choi/CoSPOT
License: http://creativecommons.org/licenses/by/4.0/
The gist: To address the sequential and evolving nature of time series, the Online Time Series Forecasting (OTSF) task has been extensively studied in multiple domains.
Terminology
Abstract
To address the sequential and evolving nature of time series, the Online Time Series Forecasting (OTSF) task has been extensively studied in multiple domains. Existing research focuses on adapting to non-stationary environments by employing memory buffer-based retrieval strategies. However, we observe that such frameworks struggle with long-term adaptation and fail to generalize to unseen patterns. To this end, we introduce CoSPOT, an LLM-based online time series forecasting framework that leverages a pre-trained LLM as the backbone online forecaster, motivated by its strong few-shot capabilities. For efficient online adaptation, CoSPOT keeps the LLM frozen and employs compositional spectral prompts grounded in frequency-domain bases to guide the model with the overall distribution of the input, thereby substantially reducing the number of parameters updated during the online phase. Specifically, CoSPOT decomposes time series into frequency bases and composes the corresponding spectral basis prompts according to their amplitudes, allowing unseen patterns to be represented as new combinations of learned basis prompts. Our extensive experiments on real-world datasets demonstrate the superiority and practicality of CoSPOT across challenging online scenarios, including extended online phases and cross-dataset settings with substantial distribution shifts. Our code is available at https://github.com/seungyoon-Choi/CoSPOT.
Sources
- Chronos-2: From Univariate to Universal Forecasting
- A decoder-only foundation model for time-series forecasting
- An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
- Time-LLM: Time Series Forecasting by Reprogramming Large Language Models
- iTransformer: Inverted Transformers Are Effective for Time Series Forecasting
- A Time Series is Worth 64 Words: Long-term Forecasting with Transformers
- Learning Fast and Slow for Online Time Series Forecasting
- LLaMA: Open and Efficient Foundation Language Models
- TimeMixer: Decomposable Multiscale Mixing for Time Series Forecasting
- TimesNet: Temporal 2D-Variation Modeling for General Time Series Analysis
- FITS: Modeling Time Series with $10k$ Parameters
Related papers
- Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
- AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
- Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
- Convergence issues in Relational Concept Analysis based on AOC-posets
- Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
- Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks