用固定随机特征提升时间序列预测效率，参数少、速度更快。
Frozen in Time: Parameter-Efficient Time Series Transformers via Reservoir-Induced Feature Expansion and Fixed Random Dynamics
- 冻结随机特征块+可训练Transformer层混合设计
- 7个基准上表现优于或媲美专用模型，计算量更低
- 适合追求高效长时序预测的研究者和工程师
Transformers虽是序列建模主流，但其二次复杂度自注意力与弱时间偏置导致长时序预测成本高且不稳定。本文提出 FreezeTST，一种轻量级混合结构，将冻结的随机特征（储层）模块与标准可训练Transformer层交替部署。冻结模块以零优化成本提供丰富非线性记忆；可训练层通过自注意力机制查询该记忆。该设计显著减少可训练参数与训练耗时，推理复杂度不变。在7个标准长时序预测基准上，FreezeTST持续匹配或超越Informer、Autoformer、PatchTST等专用变体，且所需计算资源大幅降低。结果表明，将储层原理嵌入Transformer，为高效长时序预测提供了简单而有原则的路径。
原文摘要 · Abstract (English)
Transformers are the de-facto choice for sequence modelling, yet their quadratic self-attention and weak temporal bias can make long-range forecasting both expensive and brittle. We introduce FreezeTST, a lightweight hybrid that interleaves frozen random-feature (reservoir) blocks with standard trainable Transformer layers. The frozen blocks endow the network with rich nonlinear memory at no optimisation cost; the trainable layers learn to query this memory through self-attention. The design cuts trainable parameters and also lowers wall-clock training time, while leaving inference complexity unchanged. On seven standard long-term forecasting benchmarks, FreezeTST consistently matches or surpasses specialised variants such as Informer, Autoformer, and PatchTST; with substantially lower compute. Our results show that embedding reservoir principles within Transformers offers a simple, principled route to efficient long-term time-series prediction.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。