Research map / Time series / Causal and generative time series
Reinforcement Learning Based Dynamic Model Combination for Time Series Forecasting
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use dynamic RL-based model combination instead of the single best model | Time-series forecasting ensemble · ETT | MAE: 4.59 -> 4.39 | Helped |
| Use dynamic RL-based model combination instead of the single best model | Time-series forecasting ensemble · ETT | sMAPE: 73.69 -> 72.11 | Helped |
| Use dynamic RL-based model combination instead of the single best model | Time-series forecasting ensemble · Climate | MAE: 1.67 -> 1.47 | Helped |
| Use dynamic RL-based model combination instead of the single best model | Time-series forecasting ensemble · Climate | sMAPE: 25.80 -> 25.50 | Helped |
| Use dynamic RL-based model combination instead of the single best model | Time-series forecasting ensemble · GEFCOM | MAE: 89.09 -> 88.25 | Helped |
| Use dynamic RL-based model combination instead of the single best model | Time-series forecasting ensemble · GEFCOM | sMAPE: 2.75 -> 2.73 | Helped |
| Use dynamic RL-based model combination instead of the single best model | Time-series forecasting ensemble · M4-Daily | MAE: 147.61 -> 145.78 | Helped |
| Use dynamic RL-based model combination instead of the single best model | Time-series forecasting ensemble · M4-Daily | sMAPE: 2.39 -> 2.31 | Helped |
| Add the one-hot-vertex exploration strategy (ablation) | RLMC · Climate | MAE | Helped |
| Add pretraining of the actor as a multiclass model-selection task (ablation) | RLMC · Climate | MAE | Helped |
| Add an extra replay buffer for hard, low-reward samples (ablation) | RLMC · Climate | MAE | Helped |
Closest papers by meaning
- LangTime: A Language-Guided Unified Model for Time Series Forecasting with Proximal Policy Optimization (ICML 2025 · similarity 0.84)
- SynTSBench: Rethinking Temporal Pattern Learning in Deep Learning Models for Time Series (NeurIPS 2025 · similarity 0.84)
- Time-LLM: Time Series Forecasting by Reprogramming Large Language Models (ICLR 2024 · similarity 0.84)
- ChronoEpilogi: Scalable Time Series Selection with Multiple Solutions (NeurIPS 2024 · similarity 0.83)
- Breaking Silos: Adaptive Model Fusion Unlocks Better Time Series Forecasting (ICML 2025 · similarity 0.83)
- Time-MoE: Billion-Scale Time Series Foundation Models with Mixture of Experts (ICLR 2025 · similarity 0.83)
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.