Research map / Time series / Causal and generative time series
Dynamic Dual-Feedback Conformal Inference for Time Series Forecasting
Songlin Du, Ling Luo, Uwe Aickelin
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes a dynamic dual-feedback conformal inference method for time series forecasting that adaptively adjusts prediction set thresholds under temporal dependence and distribution shift.
Problem
Uncertainty quantification for time series forecasting, where sequential data show temporal dependence and distributional shifts that break standard conformal guarantees.
Approach
Extends online conformal methods, which adjust thresholds via feedback, with a dual-feedback mechanism to adapt prediction sets dynamically; based on the abstract only, specifics are not stated.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add estimated feedback (eps=0.2) on top of actual feedback (ablation) | DDCI · GOOGL (Prophet) | MASD of interval width: 9.93 -> 6.89 | Improved |
| add estimated feedback (eps=0.2) on top of actual feedback (ablation) | DDCI · AMZN (Prophet) | MASD of interval width: 8.05 -> 5.35 | Improved |
| add actual feedback only (eps=0) versus no extra feedback (c=0) (ablation) | DDCI · GOOGL (Prophet) | MASD of interval width: 8.32 -> 9.93 | Hurt |
| add estimated feedback (ablation) | DDCI · five public datasets | coverage: at most 0.1 percentage points | No clear effect |
| increase estimated-feedback weight epsilon (ablation) | DDCI · Google stock | average interval width: wider intervals as epsilon grows | Hurt |
| estimate threshold with NexCP weights (DDCI-Nex) instead of split conformal (ablation) | DDCI · Google stock (Prophet) | average interval width: 39.20 -> 39.69 | No clear effect |
| use DDCI instead of ECI (single feedback) | online conformal · Synthetic Data 3 | coverage: 81.9 -> 89.8 | Improved |
| use DDCI instead of COP | online conformal · Google stock (Prophet) | average interval width: 49.72 -> 39.20 | Improved |
| use DDCI instead of COP | online conformal · Google stock (AR) | average interval width: 19.87 -> 19.92 | No clear effect |
| use DDCI instead of ECI | online conformal · Electricity (Prophet) | average interval width: 0.395 -> 0.392 | No clear effect |
| use DDCI instead of PID | online conformal · Electricity (Prophet) | average interval width: 0.373 -> 0.392 | Hurt |
Closest papers by meaning
- Error-quantified Conformal Inference for Time Series (ICLR 2025 · similarity 0.92)
- Conformal Prediction for Time-series Forecasting with Change Points (NeurIPS 2025 · similarity 0.89)
- Neural Conformal Control for Time Series Forecasting (AAAI 2025 · similarity 0.89)
- Conformalized Adaptive Forecasting of Heterogeneous Trajectories (ICML 2024 · similarity 0.87)
- Kernel-based Optimally Weighted Conformal Time-Series Prediction (ICLR 2025 · similarity 0.87)
- Flow-Based Conformal Predictive Distributions (NeurIPS 2026 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
