Research map / Time series / Long-term forecasting
TimeMosaic: Temporal Heterogeneity Guided Time Series Forecasting via Adaptive Granularity Patch and Segment-wise Decoding
Kuiye Ding, Fanda Fan, Chunyi Hou, Zheya Wang, Lei Wang, Zhengxin Yang, Jianfeng Zhan
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 adaptive patch embedding (ablation) | baseline patch forecaster · five datasets | MSE: 0.277 -> 0.258 | Improved |
| add adaptive patch embedding (ablation) | baseline patch forecaster · five datasets | MAE: 0.326 -> 0.302 | Improved |
| add segment-wise prompt tuning on top of APE (ablation) | TimeMosaic · five datasets | MSE: 0.258 -> 0.254 | Improved |
| add segment-wise prompt tuning on top of APE (ablation) | TimeMosaic · five datasets | MAE: 0.302 -> 0.301 | No clear effect |
| use candidate patch sizes [8, 32] instead of [8, 16, 32, 64] (ablation) | TimeMosaic · six datasets | MSE: 0.228 -> 0.228 | No clear effect |
| add more candidate patch sizes ([8, 16, 32] vs [8, 16]) (ablation) | TimeMosaic · six datasets | MSE: 0.231 -> 0.232 | Hurt |
| use segmented prediction instead of non-segmented (ablation) | TimeMosaic · five datasets, horizon 192 | MSE/MAE | Improved |
| use hyperparameter-tuned lookback up to 512 vs fixed settings | TimeMosaic · ETTm1 | MSE: 0.360 -> 0.342 | Improved |
| TimeMosaic vs DUET under hyperparameter search | TimeMosaic · ETTh2 | MSE: 0.348 vs 0.334 (DUET) | Hurt |
Later papers that compared with this method
2 later accepted papers report TimeMosaic as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- Entropy Guided Dynamic Patch Segmentation for Time Series Transformers (NeurIPS 2026 · better on 5 of 6 shared results)
- Rethinking Time Series Tokenization from a Frequency Perspective (NeurIPS 2026 · better on 2 of 3 shared results)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 2 on most shared results.
- TimeFilter — TimeFilter: Patch-Specific Spatial-Temporal Graph Filtration for Time Series Forecasting (ICML 2025) · better on 2 of 4 shared results
- PathFormer — Pathformer: Multi-scale Transformers with Adaptive Pathways for Time Series Forecasting (ICLR 2024) · better on 0 of 1 shared result
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
- Kairos: Toward Adaptive and Parameter-Efficient Time Series Foundation Models (NeurIPS 2026 · similarity 0.90)
- TimePerceiver: An Encoder-Decoder Framework for Generalized Time-Series Forecasting (NeurIPS 2025 · similarity 0.89)
- TimeBase: The Power of Minimalism in Efficient Long-term Time Series Forecasting (ICML 2025 · similarity 0.89)
- Learning Pattern-Specific Experts for Time Series Forecasting Under Patch-level Distribution Shift (NeurIPS 2025 · similarity 0.89)
- AutoTimes: Autoregressive Time Series Forecasters via Large Language Models (NeurIPS 2024 · similarity 0.89)
- Time-LLM: Time Series Forecasting by Reprogramming Large Language Models (ICLR 2024 · similarity 0.89)
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.
