TimeGuard: Channel-wise Pool Training for Backdoor Defense in Time Series Forecasting

2026-05-21

AI Summary

This work addresses the vulnerability of time series forecasting (TSF) models to backdoor attacks and the failure of existing defenses due to data entanglement and shifts in task formulation. It presents the first systematic analysis of why backdoor defenses fail in TSF and introduces TimeGuard, a novel defense framework. TimeGuard employs a channel-wise pooling training paradigm that, from early training stages, constructs a high-confidence sample pool using a time-aware criterion and dynamically expands this reliable set via a distance-regularized loss selection mechanism. This approach effectively mitigates signal dilution and loss degradation. Experiments demonstrate that TimeGuard significantly enhances robustness across diverse models and attack scenarios, reducing the poisoned MAE to 1/1.96 of the baseline while degrading clean performance by no more than 5%.

Abstract

Time Series Forecasting (TSF) plays a critical role across many domains, yet it is vulnerable to backdoor attacks. However, backdoor defenses tailored to TSF remain underexplored, due to data entanglement and task-formulation shift challenges. To fill this gap, we conduct a systematic evaluation of thirteen representative backdoor defenses across the TSF life cycle and analyze their failure modes. Our results reveal two fundamental issues: (1) data entanglement induces channel-level signal dilution, rendering sample-filtering and trigger-synthesis defenses ineffective at localizing backdoors; and (2) task-formulation shift leads to training-loss degeneration, causing poisoned and clean windows to become indistinguishable at training stages. Based on these findings, we propose a training-time backdoor defense for TSF, termed TimeGuard. Our method adopts channel-wise pool training as the core paradigm and initializes a high-confidence pool using time-aware criteria to mitigate signal dilution. Moreover, we introduce distance-regularized loss selection to progressively expand the reliable pool during training and ease loss degeneration. Extensive experiments across multiple datasets, forecasting architectures, and TSF backdoor attacks demonstrate that TimeGuard substantially improves robustness, boosting $\mathrm{MAE}_\mathrm{P}$ by $1.96\times$ over the leading baseline, while preserving clean performance within 5% $\mathrm{MAE}_\mathrm{C}$.

Problem

• backdoor defense
• time series forecasting
• data entanglement
• task-formulation shift
• signal dilution

Innovation

• channel-wise pool training
• backdoor defense
• time series forecasting
• signal dilution mitigation
• distance-regularized loss