Research map / Time series / Causal and generative time series
Supervised Dynamic Dimension Reduction with Deep Neural Network
Zhanye Luo, Yuefeng Han, Xiufan Yu
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 supervised dynamic factors (SDDP) instead of raw high-dimensional input | TCN forecaster · Climate | MAE: 3.956 -> 2.936 | Improved |
| add supervised dynamic factors (SDDP) instead of raw input | TCN forecaster · Weather | MAE: 3.991 -> 2.679 | Improved |
| replace unsupervised PCA factors with SDDP factors (ablation) | LSTM forecaster · Climate | MAE: 4.494 -> 2.946 | Improved |
| replace sdPCA with SDDP (ablation) | TCN forecaster · Climate | MAE: 3.169 -> 2.936 | Improved |
| replace sdPCA with SDDP (ablation) | TCN forecaster · Energy | MAE: 40.165 -> 41.506 | Hurt |
| add dimension reduction (SDDP, sdPCA or PCA) before TimesNet (ablation) | TimesNet forecaster · five datasets | MAE | No clear effect |
| use SDDP instead of DeepGLO | TCN forecaster · Light | MAE: 2.725 -> 2.025 | Hurt |
| increase covariate missing rate to 50% (ablation) | SDDP-TCN · Climate | MAE: 2.936 -> 3.004 | Hurt |
| add MICE imputation before SDDP (ablation) | SDDP-TCN · Climate (50% missing covariates) | MAE: 3.004 -> 2.977 | No clear effect |
| add SDDP with missing covariates instead of Vanilla | TCN forecaster · Weather (50% missing covariates) | MAE: 3.940 -> 2.662 | Improved |
| add SDDP before TimesNet with missing covariates | TimesNet forecaster · Weather (50% missing covariates) | MAE: 3.739 -> 3.712 | No clear effect |
Closest papers by meaning
- Deep Functional Factor Models: Forecasting High-Dimensional Functional Time Series via Bayesian Nonparametric Factorization (ICML 2024 · similarity 0.85)
- Time-MoE: Billion-Scale Time Series Foundation Models with Mixture of Experts (ICLR 2025 · similarity 0.85)
- DDN: Dual-domain Dynamic Normalization for Non-stationary Time Series Forecasting (NeurIPS 2024 · similarity 0.85)
- LETS Forecast: Learning Embedology for Time Series Forecasting (ICML 2025 · similarity 0.85)
- TStruct: Learning Shared Temporal Structures for Long-Term Time-Series Forecasting (NeurIPS 2026 · similarity 0.84)
- Structured Matrix Basis for Multivariate Time Series Forecasting with Interpretable Dynamics (NeurIPS 2024 · 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.
