Few-Shot Forecasting of Time-Series with Heterogeneous Channels
This addresses the challenge of data scarcity in time-series forecasting for domains requiring efficient learning across tasks, though it is incremental as it extends prior work on heterogeneous attributes to time-series.
The paper tackles the problem of few-shot forecasting for multivariate time-series with heterogeneous channels, which existing methods cannot handle due to differences in channels and the forecasting task. It introduces a model using permutation-invariant deep set-blocks with temporal embedding, and experiments on a new meta-dataset of 40 datasets show it outperforms baselines by providing better generalization.
Learning complex time series forecasting models usually requires a large amount of data, as each model is trained from scratch for each task/data set. Leveraging learning experience with similar datasets is a well-established technique for classification problems called few-shot classification. However, existing approaches cannot be applied to time-series forecasting because i) multivariate time-series datasets have different channels and ii) forecasting is principally different from classification. In this paper we formalize the problem of few-shot forecasting of time-series with heterogeneous channels for the first time. Extending recent work on heterogeneous attributes in vector data, we develop a model composed of permutation-invariant deep set-blocks which incorporate a temporal embedding. We assemble the first meta-dataset of 40 multivariate time-series datasets and show through experiments that our model provides a good generalization, outperforming baselines carried over from simpler scenarios that either fail to learn across tasks or miss temporal information.