# TimeMosaic: Temporal Heterogeneity Guided Time Series Forecasting via Adaptive Granularity Patch and Segment-wise Decoding

arXiv: https://arxiv.org/abs/2509.19406 · HTML: https://mlsanity.com/paper/2509.19406/

## Key facts

- Cited by 12 papers in the map (data through 14 Oct 2026).
- Builds on 33 of its 55 references that are in the map.
- First posted on arXiv 23 Sep 2025.

## Builds on (most cited first, 30 of 33)

- [PyTorch: An Imperative Style, High-Performance Deep Learning Library](https://mlsanity.com/paper/1912.01703/) (2019)
- [The Power of Scale for Parameter-Efficient Prompt Tuning](https://mlsanity.com/paper/2104.08691/) (2021)
- [Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting](https://mlsanity.com/paper/2012.07436/) (2020)
- [Modeling Long- and Short-Term Temporal Patterns with Deep Neural Networks](https://mlsanity.com/paper/1703.07015/) (2017)
- [An Overview of Multi-Task Learning in Deep Neural Networks](https://mlsanity.com/paper/1706.05098/) (2017)
- [GPT Understands, Too](https://mlsanity.com/paper/2103.10385/) (2021)
- [One Fits All:Power General Time Series Analysis by Pretrained LM](https://mlsanity.com/paper/2302.11939/) (2023)
- [Unified Training of Universal Time Series Forecasting Transformers](https://mlsanity.com/paper/2402.02592/) (2024)
- [Sequence Level Training with Recurrent Neural Networks](https://mlsanity.com/paper/1511.06732/) (2015)
- [MOMENT: A Family of Open Time-series Foundation Models](https://mlsanity.com/paper/2402.03885/) (2024)
- [Multi-Task Learning with Deep Neural Networks: A Survey](https://mlsanity.com/paper/2009.09796/) (2020)
- [Time-MoE: Billion-Scale Time Series Foundation Models with Mixture of Experts](https://mlsanity.com/paper/2409.16040/) (2024)
- [WeatherBench: A benchmark dataset for data-driven weather forecasting](https://mlsanity.com/paper/2002.00469/) (2020)
- [ToolQA: A Dataset for LLM Question Answering with External Tools](https://mlsanity.com/paper/2306.13304/) (2023)
- [Sentiment Analysis in the Era of Large Language Models: A Reality Check](https://mlsanity.com/paper/2305.15005/) (2023)
- [TFB: Towards Comprehensive and Fair Benchmarking of Time Series Forecasting Methods](https://mlsanity.com/paper/2403.20150/) (2024)
- [Sundial: A Family of Highly Capable Time Series Foundation Models](https://mlsanity.com/paper/2502.00816/) (2025)
- [Less Is More: Fast Multivariate Time Series Forecasting with Light Sampling-oriented MLP Structures](https://mlsanity.com/paper/2207.01186/) (2022)
- [TimeMixer++: A General Time Series Pattern Machine for Universal Predictive Analysis](https://mlsanity.com/paper/2410.16032/) (2024)
- [FreDF: Learning to Forecast in the Frequency Domain](https://mlsanity.com/paper/2402.02399/) (2024)
- [Towards Accurate Image Coding: Improved Autoregressive Image Generation with Dynamic Vector Quantization](https://mlsanity.com/paper/2305.11718/) (2023)
- [DSformer: A Double Sampling Transformer for Multivariate Time Series Long-term Prediction](https://mlsanity.com/paper/2308.03274/) (2023)
- [DAG: A Dual Correlation Network for Time Series Forecasting with Exogenous Variables](https://mlsanity.com/paper/2509.14933/) (2025)
- [BuildingsBench: A Large-Scale Dataset of 900K Buildings and Benchmark for Short-Term Load Forecasting](https://mlsanity.com/paper/2307.00142/) (2023)
- [SaDI: A Self-adaptive Decomposed Interpretable Framework for Electric Load Forecasting under Extreme Events](https://mlsanity.com/paper/2306.08299/) (2023)
- [GinAR: An End-To-End Multivariate Time Series Forecasting Model Suitable for Variable Missing](https://mlsanity.com/paper/2405.11333/) (2024)
- [BLAST: Balanced Sampling Time Series Corpus for Universal Forecasting Models](https://mlsanity.com/paper/2505.17871/) (2025)
- [LSR-IGRU: Stock Trend Prediction Based on Long Short-Term Relationships and Improved GRU](https://mlsanity.com/paper/2409.08282/) (2024)
- [DualSG: A Dual-Stream Explicit Semantic-Guided Multivariate Time Series Forecasting Framework](https://mlsanity.com/paper/2507.21830/) (2025)
- [What is the best RNN-cell structure to forecast each time series behavior?](https://mlsanity.com/paper/2203.07844/) (2022)

## Built on by (most cited first, 12 of 12)

- [Entropy Guided Dynamic Patch Segmentation for Time Series Transformers](https://mlsanity.com/paper/2509.26157/) (2025)
- [TiWeaver: Unified Temporal Dynamics Modeling via Contextual Patching](https://mlsanity.com/paper/2606.03121/) (2026)
- [What If We Let Forecasting Forget? A Sparse Bottleneck for Cross-Variable Dependencies](https://mlsanity.com/paper/2605.08289/) (2026)
- [Multivariate Time Series Forecasting needs Cross Variable Loss](https://mlsanity.com/paper/2608.05742/) (2026)
- [QiYao-M: Multimodal Time Series Foundation Model with Role-Aware Modeling of Endogenous and Exogenous Modalities](https://mlsanity.com/paper/2609.34842/) (2026)
- [KAIROS: Unified Training for Universal Non-Autoregressive Time Series Forecasting](https://mlsanity.com/paper/2510.02084/) (2025)
- [We Need a More Robust Classifier: Dual Causal Learning Empowers Domain-Incremental Time Series Classification](https://mlsanity.com/paper/2601.10312/) (2026)
- [Adaptive Patching Is Harder Than It Looks For Time-Series Forecasting](https://mlsanity.com/paper/2606.04074/) (2026)
- [Scale-Aware Pretraining of Time Series Foundation Models via Multi-Patch Token Alignment and Hybrid Masking](https://mlsanity.com/paper/2608.20005/) (2026)
- [CoRe: Coherence and Relational Alignment for Multivariate Time Series Forecasting](https://mlsanity.com/paper/2609.19670/) (2026)
- [DP-Rec: Towards Dynamic Patching for Efficient Long-Sequence Recommendation](https://mlsanity.com/paper/2609.32215/) (2026)
- [HALO: Enhancing Time Series Generation via Hyperspherical Latents and Masked AutoregRessive Modeling](https://mlsanity.com/paper/2609.34511/) (2026)

## Abstract (arXiv)

Multivariate time series forecasting is essential in domains such as finance, transportation, climate, and energy. However, existing patch-based methods typically adopt fixed-length segmentation, overlooking the heterogeneity of local temporal dynamics and the decoding heterogeneity of forecasting. Such designs lose details in information-dense regions, introduce redundancy in stable segments, and fail to capture the distinct complexities of short-term and long-term horizons. We propose TimeMosaic, a forecasting framework that aims to address temporal heterogeneity. TimeMosaic employs adaptive patch embedding to dynamically adjust granularity according to local information density, balancing motif reuse with structural clarity while preserving temporal continuity. In addition, it introduces segment-wise decoding that treats each prediction horizon as a related subtask and adapts to horizon-specific difficulty and information requirements, rather than applying a single uniform decoder. Extensive evaluations on benchmark datasets demonstrate that TimeMosaic delivers consistent improvements over existing methods, and our model trained on the large-scale corpus with 321 billion observations achieves performance competitive with state-of-the-art TSFMs.