Research map / Probabilistic and scientific modeling / Dynamical systems
DRL-STAF: A Deep Reinforcement Learning Framework for State-Aware Forecasting of Complex Multivariate Hidden Markov Processes
Manrui Jiang, Jingru Huang, Yong Chen, Chen Zhang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| add hidden-state modeling (DRL-S1 over DL-F) (ablation) | DL-F forecaster · Simulated 3 variables | MAE: 0.2599 -> 0.0956 | Improved |
| add hidden-state modeling (DRL-S1 over DL-F) (ablation) | DL-F forecaster · Simulated 10 variables | MAE: 0.3646 -> 0.1129 | Improved |
| add hidden-state modeling (DRL-S1 over DL-F) (ablation) | DL-F forecaster · Exchange | MAE: 3.5285 -> 1.8412 | Improved |
| add hidden-state modeling (DRL-S1 over DL-F) (ablation) | DL-F forecaster · Traffic | MAE: 1.8387 -> 1.5686 | Improved |
| add second-stage cross-variable interaction training (ablation) | DRL-S1 · Simulated 3 variables | MAE: 0.0956 -> 0.0889 | Improved |
| add second-stage cross-variable interaction training (ablation) | DRL-S1 · Simulated 10 variables | MAE: 0.1129 -> 0.1090 | Improved |
| add second-stage cross-variable interaction training (ablation) | DRL-S1 · Exchange | MAE: 1.8412 -> 1.7249 | Improved |
| add second-stage cross-variable interaction training (ablation) | DRL-S1 · Traffic | MAE: 1.5686 -> 1.5193 | Improved |
| add second-stage cross-variable interaction training (ablation) | DRL-S1 · Simulated 3 variables | accuracy: 97.53% -> 98.17% | Improved |
| add second-stage cross-variable interaction training (ablation) | DRL-S1 · SMachine | accuracy: 85.80% -> 81.73% | Hurt |
| add second-stage cross-variable interaction training (ablation) | DRL-S1 · SMachine | F1: 72.85% -> 77.50% | Improved |
| add second-stage cross-variable interaction training (ablation) | DRL-S1 · Simulated 10 variables | recall: 93.23% -> 91.89% | Hurt |
Closest papers by meaning
- Hierarchical Deep Generative Models for Multi-Rate Multivariate Time Series (ICML 2018 · similarity 0.86)
- Learning Markov State Abstractions for Deep Reinforcement Learning (NeurIPS 2021 · similarity 0.86)
- LangTime: A Language-Guided Unified Model for Time Series Forecasting with Proximal Policy Optimization (ICML 2025 · similarity 0.85)
- Reinforcement Learning with Imperfect Transition Predictions: A Bellman-Jensen Approach (NeurIPS 2025 · similarity 0.84)
- Bridging State and History Representations: Understanding Self-Predictive RL (ICLR 2024 · similarity 0.84)
- Rethinking Post-Training Recipes for Multimodal Time-Series Forecasting (NeurIPS 2026 · similarity 0.84)
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.
