aiqbench analysis
Trio introduces a novel architecture that explicitly integrates temporal, spatial, and sample-aware attention mechanisms to address the limitations of standard tabular priors in multivariate time-series forecasting. By leveraging a Time-Series Structural Causal Model generator to create synthetic tasks with dynamic lags and distributional drift, the method demonstrates improved performance on industrial and public benchmarks. This work highlights a promising direction for organizing and reusing historical input-output examples to enhance model generalization in complex forecasting scenarios.
Trio: Learning Time-Series Forecasting with Temporal-Spatial-Sample Attention and Structural Causal Priors
arXiv:2606.07291v1 Announce Type: new Abstract: Multivariate time-series forecasting requires models to reason over temporal dynamics, cross-variable dependencies, and historical input-output correspondences. Recent Prior-Data Fitted Networks (PFNs) suggest that synthetic tasks can be useful for learning transferable inference behavior. However, directly transferring this paradigm to time-series forecasting remains difficult, since temporal order, dynamic lags, and recurring historical patterns are not naturally captured by ordinary tabular priors. Motivated by this observation, we propose Trio, a sample-aware time-series forecasting architecture based on Temporal-Spatial-Sample attention. Temporal attention captures within-window dynamics, spatial attention models inter-variable dependencies, and sample attention retrieves relevant historical lookback-future pairs to guide the current prediction. Rather than claiming a fully general PFN-style forecaster, our goal is to study how historical input-output examples can be explicitly organized and reused within a forecasting model. We further introduce a Time-Series Structural Causal Model (TS-SCM) generator to create structured synthetic forecasting tasks with dynamic lags, cross-variable interactions, noise, feedback, and distributional drift. Experiments on synthetic, industrial, and public benchmarks show that the proposed architecture improves forecasting performance. Exploratory zero-shot experiments further suggest that TS-SCM-generated tasks may provide useful structural priors, while fully general PFN-style time-series forecasting remains an open problem.
Abstract:Multivariate time-series forecasting requires models to reason over temporal dynamics, cross-variable dependencies, and historical input-output correspondences. Recent Prior-Data Fitted Networks (PFNs) suggest that synthetic tasks can be useful for learning transferable inference behavior. However, directly transferring this paradigm to time-series forecasting remains difficult, since temporal order, dynamic lags, and recurring historical patterns are not naturally captured by ordinary tabular priors. Motivated by this observation, we propose Trio, a sample-aware time-series forecasting architecture based on Temporal-Spatial-Sample attention. Temporal attention captures within-window dynamics, spatial attention models inter-variable dependencies, and sample attention retrieves relevant historical lookback-future pairs to guide the current prediction. Rather than claiming a fully general PFN-style forecaster, our goal is to study how historical input-output examples can be explicitly organized and reused within a forecasting model. We further introduce a Time-Series Structural Causal Model (TS-SCM) generator to create structured synthetic forecasting tasks with dynamic lags, cross-variable interactions, noise, feedback, and distributional drift. Experiments on synthetic, industrial, and public benchmarks show that the proposed architecture improves forecasting performance. Exploratory zero-shot experiments further suggest that TS-SCM-generated tasks may provide useful structural priors, while fully general PFN-style time-series forecasting remains an open problem.
Machine Learning (cs.LG)
arXiv:2606.07291 [cs.LG]
(or arXiv:2606.07291v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2606.07291 arXiv-issued DOI via DataCite (pending registration)
Full text under the publisher's open licence · 208 words
Related coverage
Modeling spatio-temporal locality in multi-step forecasting of geo-referenced time series
arXiv · Aug 27, 2026
VFEM: Visual Feature Empowered Multivariate Time Series Forecasting with Cross-Modal Fusion
arXiv · Jun 9, 2026
In-Context Inpainting for Time Series Forecasting
arXiv · Aug 27, 2026
WinoTS: Wavelet-based Self-Distillation for Time Series Models
arXiv · Oct 1, 2026