CARE: A Lightweight Plug-in Gated Correction and Uncertainty-aware Module for Long-term Time Series Forecasting

Key points

• Multivariate long-horizon forecasting is critical to electricity load scheduling and traffic flow management, and to financial risk control.1
• Operating in parallel with the base model, CARE resamples historical context to match the forecast horizon, learns residual correction patterns from this aligned history, and applies scale-aware bounded updates modulated by per-coordinate sigmoid risk gates.1
• A multi-objective loss jointly optimizes forecast accuracy, residual tracking, risk alignment, and base-model anchoring.1
• Its risk gates reliably identify high-error regions: on Weather, the highest-gate tertile exhibits nearly four times the error of the lowest-gate tertile, offering planners an interpretable per-step trust signal.1

Sources (1)

• [1]CARE: A Lightweight Plug-in Gated Correction and Uncertainty-aware Module for Long-term Time Series Forecasting
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 8, 03:23 AM

  We present CARE (Corrective branch with Aligned context and Relative-error Estimation), a lightweight plug-in that enhances any deterministic forecaster without architectural redesign.

  Multivariate long-horizon forecasting is critical to electricity load scheduling and traffic flow management, and to financial risk control.

Extractive summary: sentences quoted from the sources.