UPLOTS: A Unified Pretrained Language Model for Constrained Time-series Generation

2026-06-09

AI Summary

This work addresses the limitations of existing time series generation methods, which are typically tailored to individual datasets, lack cross-domain reusability, and overlook shared temporal structures. To overcome these challenges, we propose UPLOTS, a unified framework built upon a pretrained Transformer backbone that enables multi-domain, on-demand, and pattern-controllable generation through learnable constraint prompts. By integrating dynamic multi-dataset loss reweighting with a prompt-to-pattern mapping mechanism, UPLOTS internalizes diverse temporal structures during training and supports conditional generation at inference. Experimental results demonstrate that UPLOTS effectively accommodates composite constraints and enhances downstream forecasting tasks across four real-world benchmarks, significantly improving data augmentation performance in data-scarce scenarios.

Abstract

In time-series generation, existing approaches typically handcraft ortrain a separate model for each dataset, which hinders their scalability and fails to leverage shared temporal structures across domains. To address this fragmentation, we propose UPLOTS, a Unified, Prompt-guided Language model framework fOr constrained Time-Series Generation across diverse domains. Instead of building task-specific models, UPLOTS leverages a single pre-trained transformer backbone guided by learned constraint prompts, enabling on-demand generation with precise pattern control. One key innovation is our dynamic multi-dataset loss re-weighting and prompt-to-pattern mapping, which allows UPLOTS to internalize diverse temporal structures during training and conditionally generate them at inference. We evaluate UPLOTS on four real-world benchmarks and multiple constraint settings, including peak-period, calendar, load-level, and volatility patterns. Additional held-out constraint-combination and downstream forecasting experiments further demonstrate that UPLOTS generalizes beyond the original peak-pattern setting and improves data augmentation under scarce real-data regimes. Our code and baselines are available at anonymous github repo: https://anonymous.4open.science/r/UPLOTS-6C36.

Problem

• time-series generation
• model fragmentation
• cross-domain scalability
• temporal structures
• constrained generation

Innovation

• constrained time-series generation
• unified pretrained language model
• prompt-guided generation
• dynamic loss re-weighting
• temporal pattern control