Get Started
Home
Topics
Search
Library
Research questionIn multivariate forecasting, how can augmentation vary samples without breaking input–future temporal coherence?Forecasting augmentations can disrupt temporal relationships between the look-back window and its continuous future target. The central difficulty is adding useful variation without producing synthetic examples that no longer represent a coherent forecasting task.
AI
Evaluation & Benchmarks
Machine Learning
Statistical Machine Learning
Latest papersRecent research connected to this question, newest first.Sliding-Window Reordering with Overlap Averaging: A Simple Time-Domain Augmentation for Multivariate ForecastingThe source concerns a model-agnostic time-domain augmentation for multivariate forecasting. Evidence covers nine long-term forecasting benchmarks across five backbone families and four short-term traffic benchmarks using PatchTST; transfer to univariate and multivariate classification is also reported.research paper · Sep 3, 2026
Related questions
How can multimodal time-series forecasting align sparse text with heterogeneous temporal dynamics?When should self-supervised pretraining pool dependent augmentations rather than partition data into independent subsets?How can online time-series forecasters adapt to evolving, unseen patterns without costly continual parameter updates?How can multivariate time-series imputers recover structured gaps under high missingness?