Exogenous Dropout: A Simple, Strong Baseline for Corruption-Robust Time Series Forecasting with Covariates

2026-07-05

AI Summary

Time series forecasting with exogenous covariates often suffers significant performance degradation—and consequently lacks deployment robustness—when those covariates are corrupted by noise, temporal misalignment, or missing values. This work proposes Exogenous Dropout, a simple yet effective training strategy that randomly zeros out entire exogenous channels during training, thereby enhancing model robustness to covariate corruption without requiring architectural complexity, while preserving predictive accuracy on clean data. The method is model-agnostic and is validated through integration with dual-correlation networks, learnable gating, residual fallback mechanisms, and channel-wise FiLM modulation. Extensive experiments across electricity price, hydrological, and meteorological datasets demonstrate substantial improvements over existing approaches, including BoundEx—a method explicitly designed for robustness—thereby establishing, for the first time, that explicitly bounded architectures are not a prerequisite for achieving strong robustness.

Abstract

Time series forecasters that use exogenous covariates are fragile in deployment: when those covariates are noised, temporally misaligned, or missing, strong exogenous-fusion and exogenous-adapted models can degrade far above the endogenous-only floor. We study whether such robustness requires specialized architectures, or whether it can be obtained through a simple training intervention. We propose exogenous dropout, a model-agnostic method that randomly zeros whole exogenous channels during training. Across electricity-price forecasting, reservoir hydrology, and meteorology, exogenous dropout substantially improves robustness under Gaussian noise, temporal misalignment, and fully missing channels, while preserving clean accuracy. Applied to a dual-correlation network, it yields the most robust model in our experiments, outperforming a deliberately strong bounded architectural foil, BoundEx, which combines a learnable gate, a fallback residual to the endogenous backbone, and per-channel exogenous FiLM modulation. Architecture-by-dropout ablations, gate-behavior diagnostics, and a representation-level bound show that explicit architectural boundedness is not necessary for this robustness: an unbounded model trained with exogenous dropout is more robust than the bounded model in every domain. We release a corruption-robustness benchmark and recommend exogenous dropout as a simple, strong baseline for future work on time series forecasting with covariates.

Problem

• time series forecasting
• exogenous covariates
• corruption robustness
• model fragility
• robustness benchmark

Innovation

• Exogenous Dropout
• Corruption-Robust Forecasting
• Time Series with Covariates
• Model-Agnostic Training
• Robustness Benchmark