Research map / Time series / Causal and generative time series
Beyond MSE: Ordinal Cross-Entropy for Probabilistic Time Series Forecasting
Jieting Wang, Huimei Shi, Feijiang Li, Xiaolei Shang
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 |
|---|---|---|---|
| replace MSE regression with ordinal cross-entropy classification (OCE-TS) | DLinear · ETTm2 (H=96) | MSE: 0.167 -> 0.118 | Improved |
| replace MSE regression with OCE-TS | DLinear · Exchange (H=720) | MSE: 0.643 -> 0.521 | Improved |
| replace MSE regression with OCE-TS | DLinear · ILI (H=24) | MSE: 2.215 -> 0.570 | Improved |
| replace MSE regression with OCE-TS | DLinear · ETTh1 (H=720) | MSE: 0.472 -> 0.547 | Hurt |
| replace MSE regression with OCE-TS | DLinear · ETTm1 (H=720) | MSE: 0.425 -> 0.479 | Hurt |
| OCE-TS vs strongest baseline TimeBridge | TimeBridge · 7 datasets | MSE: ETTh1 H=720 0.447 -> 0.547; Weather H=720 0.307 -> 0.225 | Mixed |
| use OCE loss instead of plain CE (ablation) | OCE-TS · Weather | MSE/MAE | Improved |
| choose truncated Gaussian over Student-t or Laplace (ablation) | OCE-TS · ETTh1 | MSE/MAE | Improved |
| choose truncated Gaussian over Student-t or Laplace (ablation) | OCE-TS · ETTh2, ETTm1, ETTm2 | MSE/MAE | Mixed |
| increase sigma to 1 (ablation) | OCE-TS · ETTm1 | MSE | Hurt |
| vary number of bins (30 to 100) (ablation) | OCE-TS · ETTm1, ETTh2 | MSE | No clear effect |
| increase lookback window to 336 or more (ablation) | OCE-TS · ETTm1 (H=720) | MSE | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- TimeBridge — TimeBridge: Non-Stationarity Matters for Long-term Time Series Forecasting (ICML 2025) · better on 12 of 14 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- OLinear: A Linear Model for Time Series Forecasting in Orthogonally Transformed Domain (NeurIPS 2025 · similarity 0.85)
- Multivariate Probabilistic Time Series Forecasting with Correlated Errors (NeurIPS 2024 · similarity 0.85)
- SIN: Selective and Interpretable Normalization for Long-Term Time Series Forecasting (ICML 2024 · similarity 0.85)
- ProbTS: Benchmarking Point and Distributional Forecasting across Diverse Prediction Horizons (NeurIPS 2024 · similarity 0.84)
- Time-Series Forecasting for Out-of-Distribution Generalization Using Invariant Learning (ICML 2024 · similarity 0.84)
- Selective Learning for Deep Time Series Forecasting (NeurIPS 2025 · 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.
