TimeOmni-VL: Unified Models for Time Series Understanding and Generation
This work addresses a fundamental gap in time series research by bridging generation and understanding, potentially impacting fields like finance, healthcare, and IoT through more accurate and interpretable models.
The paper tackles the divide between numerical generation and semantic understanding in time series modeling by proposing TimeOmni-VL, a vision-centric framework that unifies these tasks through fidelity-preserving bidirectional mapping and understanding-guided generation, achieving significant improvements in both semantic understanding and numerical precision.
Recent time series modeling faces a sharp divide between numerical generation and semantic understanding, with research showing that generation models often rely on superficial pattern matching, while understanding-oriented models struggle with high-fidelity numerical output. Although unified multimodal models (UMMs) have bridged this gap in vision, their potential for time series remains untapped. We propose TimeOmni-VL, the first vision-centric framework that unifies time series understanding and generation through two key innovations: (1) Fidelity-preserving bidirectional mapping between time series and images (Bi-TSI), which advances Time Series-to-Image (TS2I) and Image-to-Time Series (I2TS) conversions to ensure near-lossless transformations. (2) Understanding-guided generation. We introduce TSUMM-Suite, a novel dataset consists of six understanding tasks rooted in time series analytics that are coupled with two generation tasks. With a calibrated Chain-of-Thought, TimeOmni-VL is the first to leverage time series understanding as an explicit control signal for high-fidelity generation. Experiments confirm that this unified approach significantly improves both semantic understanding and numerical precision, establishing a new frontier for multimodal time series modeling.