方法对比

	Time-MoE：面向通用时间序列的混合专家模型	Chronos：一种用于时间序列预测的标记化基础模型	专家混合模型	TimesFM：面向时间序列预测的仅解码器基础模型
领域	深度学习	深度学习	深度学习	深度学习
方法族	Machine learning	Machine learning	Machine learning	Machine learning
起源年份	2024	2024	2017	2024
提出者	Xiaoming Shi et al.	Abdul Fatir Ansari et al. (Amazon)	Shazeer, N. et al.	Abhimanyu Das et al. (Google)
类型	Sparse mixture-of-experts autoregressive foundation model	Pre-trained language-model-based time-series forecaster	Sparse neural network architecture (conditional computation)	Pre-trained decoder-only transformer for zero-shot time-series forecasting
开创性文献	Shi, X., Wang, S., Nie, Y., Li, D., Ye, Z., Wen, Q., & Jin, M. (2024). Time-MoE: Billion-scale time series foundation models with mixture of experts. ICLR 2025.	Ansari, A. F., Stella, L., Turkmen, C., Zhang, X., Mercado, P., Shen, H., et al. (2024). Chronos: Learning the language of time series. Transactions on Machine Learning Research.	Shazeer, N. et al. (2017). Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layer. ICLR. arXiv:1701.06538	Das, A., Kong, W., Sen, R., & Zhou, Y. (2024). A decoder-only foundation model for time-series forecasting. ICML.
别名	Time Mixture-of-Experts, Time-MoE Foundation Model, Sparse Time-Series Transformer, Zaman Karışık Uzmanlar Modeli	Chronos Forecasting Model, Amazon Chronos, Tokenized Time-Series LLM, Kronos Zaman Serisi Modeli	Uzman Karışımı (Mixture of Experts — MoE), uzman karışımı, MoE, sparse mixture of experts	Time-series Foundation Model, Google TimesFM, TimesFM forecaster, Zaman Serisi Temel Modeli
相关	3	2	3	3
摘要	Time-MoE is a billion-scale autoregressive foundation model for universal time-series forecasting, introduced by Shi et al. in 2024 and accepted at ICLR 2025. It combines a decoder-only transformer architecture with sparse Mixture-of-Experts (MoE) feed-forward layers, enabling the model to scale to billions of parameters while activating only a small subset of expert networks per token—dramatically increasing capacity without proportional compute cost.	Chronos is a family of pre-trained probabilistic forecasting models introduced by Ansari et al. at Amazon in 2024. It adapts the language-model paradigm to time series by quantizing continuous values into discrete tokens, enabling a standard transformer to be trained on a large heterogeneous corpus of time-series data. The result is a zero-shot forecasting model that generalizes across domains without requiring dataset-specific retraining.	Mixture of Experts (MoE) is a sparse neural-network architecture, introduced by Shazeer and colleagues in 2017 with the sparsely-gated MoE layer, in which only a subset of expert sub-networks is activated for each input. As seen in models such as Switch Transformer and Mixtral, it holds computation cost fixed even as the total parameter count grows.	TimesFM is a pre-trained foundation model for univariate time-series forecasting introduced by Abhimanyu Das, Weihao Kong, Rajat Sen, and Yichen Zhou from Google in 2024. The model adopts a decoder-only transformer architecture, similar in spirit to large language models, and is trained on a large corpus of real-world and synthetic time-series data. Its central innovation is the ability to perform accurate zero-shot forecasting across diverse domains without task-specific fine-tuning.