mlsanityData to 7 Oct 2026

AME-TS: Anchored Mixture-of-Experts for Time Series Forecasting

arXiv 2605.25166 ↗First posted 24 May 2026cs.LG cs.AIcited by 1

On the map: Mixture-of-experts → MoE for time series and graphs

Abstract (arXiv)

Time series forecasting models are increasingly scaled through large Transformer backbones, yet most existing approaches process all series through a shared dense computation path despite substantial heterogeneity in temporal structure. Mixture-of-Experts (MoE) offers a natural alternative by enabling conditional computation, but standard MoE routing leaves expert specialization weakly identified and often unstable during downstream adaptation. We propose AME-TS, a structure-guided sparse time series foundation model that aligns expert routing with interpretable temporal structure. AME-TS first uses a lightweight regime predictor to estimate series-level descriptors, including forecastability, seasonality, trend, and sparsity, and maps them to a soft structural prior over experts. This series-level prior guides token-level routing during training, encouraging structure-aligned specialization. On the GIFT-Eval benchmark, AME-TS delivers a strong accuracy-efficiency tradeoff across model scales: it substantially outperforms existing time series foundation models at small model scales and remains competitive with the strongest models at larger scales, while activating substantially fewer parameters through sparse routing. We further show that AME-TS learns more interpretable routing geometry and substantially more stable expert specialization than standard MoE during fine-tuning on the M5 dataset. These results suggest that structure-aware routing is an effective and reliable way to realize the benefits of sparse expert models for time series forecasting.

Key facts. Cited by 1 paper in the map (data through 7 Oct 2026). Placed in MoE for time series and graphs, one of 12 directions in Mixture-of-experts. Builds on 30 of its 37 references that are in the map. First posted on arXiv 24 May 2026.

Builds on

30 of 37 references.

Closest ancestors · 2

Same direction, newest first

  1. 2024Moirai-MoE: Empowering Time Series Foundation Models with Sparse Mixture of Experts14 Oct
  2. 2024Time-MoE: Billion-Scale Time Series Foundation Models with Mixture of Experts24 Sep

Other references · 28

  1. 2021Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsitycited by 3,014
  2. 2020Fourier Neural Operator for Parametric Partial Differential Equationscited by 2,806
  3. 2022A Time Series is Worth 64 Words: Long-term Forecasting with Transformerscited by 1,981
  4. 2024Mixtral of Expertscited by 1,785
  5. 2023iTransformer: Inverted Transformers Are Effective for Time Series Forecastingcited by 906
  6. 2024DeepSeekMoE: Towards Ultimate Expert Specialization in Mixture-of-Experts Language Modelscited by 834
  7. 2024Chronos: Learning the Language of Time Seriescited by 712
  8. 2023A decoder-only foundation model for time-series forecastingcited by 650
  9. 2024Unified Training of Universal Time Series Forecasting Transformerscited by 588
  10. 2024A Survey on Mixture of Experts in Large Language Modelscited by 233
  11. 2019Towards Physics-informed Deep Learning for Turbulent Flow Predictioncited by 203
  12. 2024GIFT-Eval: A Benchmark For General Time Series Forecasting Model Evaluationcited by 188
16 more
  1. 2025Sundial: A Family of Highly Capable Time Series Foundation Modelscited by 168
  2. 2025Chronos-2: From Univariate to Universal Forecastingcited by 158
  3. 2020Deep Learning for Time Series Forecasting: Tutorial and Literature Surveycited by 117
  4. 2025TiRex: Zero-Shot Forecasting Across Long and Short Horizons with Enhanced In-Context Learningcited by 107
  5. 2024MoVA: Adapting Mixture of Vision Experts to Multimodal Contextcited by 82
  6. 2021Time Series Prediction using Deep Learning Methods in Healthcarecited by 71
  7. 2023Mixture-of-Experts Meets Instruction Tuning:A Winning Combination for Large Language Modelscited by 65
  8. 2025Moirai 2.0: When Less Is More for Time Series Forecastingcited by 59
  9. 2025fev-bench: A Realistic Benchmark for Time Series Forecastingcited by 52
  10. 2024MoME: Mixture of Multimodal Experts for Generalist Multimodal Large Language Modelscited by 45
  11. 2025Advancing Expert Specialization for Better MoEcited by 37
  12. 2024Multi-Head Mixture-of-Expertscited by 30
  13. 2025Routing Matters in MoE: Scaling Diffusion Transformers with Explicit Routing Guidancecited by 23
  14. 2025Conversational Time Series Foundation Models: Towards Explainable and Effective Forecastingcited by 8
  15. 2025Guiding the Experts: Semantic Priors for Efficient and Focused MoE Routingcited by 1
  16. 2025Time Series Forecastability Measurescited by 1

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026Fuzzy-MoE: Interpretable Regime-Conditioned Expert Routing for Non-Stationary Multivariate Time Series Forecastingcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Seg-MoE: Multi-Resolution Segment-wise Mixture-of-Experts for Time Series Forecasting Transformers
  2. 2026MoHETS: Long-term Time Series Forecasting with Mixture-of-Heterogeneous-Experts
  3. 2026FAME: Forecastability-Aware Mixture of Experts for Heterogeneous Time Series Forecasting
  4. 2025Time Tracker: Mixture-of-Experts-Enhanced Foundation Time Series Forecasting Model with Decoupled Training Pipelines
  5. 2026WaveMoE: A Wavelet-Enhanced Mixture-of-Experts Foundation Model for Time Series Forecasting
  6. 2025MoFE-Time: Mixture of Frequency Domain Experts for Time-Series Forecasting Models
  7. 2023Mixture-of-Linear-Experts for Long-term Time Series Forecasting
  8. 2026Dynamic TMoE: A Drift-Aware Dynamic Mixture of Experts Framework for Non-Stationary Time Series Forecasting