arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2606.08896cs.AI
收藏

FAME: 面向异构时间序列预测的可预测性感知专家混合模型

FAME: Forecastability-Aware Mixture of Experts for Heterogeneous Time Series Forecasting

Sun Yat-sen University(中山大学);Guangdong Key Laboratory of Information Security Technology(广东省信息安全技术重点实验室);Ministry of Education Key Laboratory of Machine Intelligence and Advanced Computing(教育部机器智能与先进计算重点实验室)

Qianyang Li, Xingjun Zhang, Shaoxun Wang, Tao Peng, Jia Wei

AI总结:

针对大规模异构时间序列预测中单一模型性能不足的问题,提出可预测性感知的稀疏专家混合框架FAME,通过多维可预测性指纹和成本感知路由,在工业数据集上实现12.4%的MSE降低。

摘要:

大规模零售和工业预测系统包含许多异构时间序列,其生命周期、稀疏性、波动性、季节性、频谱模式和上下文敏感性差异很大。单一预测模型很少能在所有情况下表现良好,而密集集成会增加推理成本并提供有限的专家适用性洞察。本文研究可预测性感知的专家路由:学习数据特征如何决定预测专家的适用性。我们提出\method{},一个稀疏专家混合框架,用多维可预测性指纹表示每个序列,从验证性能中挖掘专家适用性目标,并训练一个成本感知的稀疏路由器,为每个序列激活少量预算的专家集。使用山东新北洋(SNBC)的生产规模自动售货机销售数据集(其中预测组件已集成到补货计划管道中)以及公共零售基准,我们表明专家适用性在不同数据情况下系统性地变化。在拥有5000+台机器和6000万+交易的工业数据集上,\method{} Top-2相比最强单一专家LightGBM降低了12.4%的MSE,同时平均每个序列执行1.92个专家。部署的组件产生需求预测,而库存导向的收益通过离线回放模拟器在固定补货策略下估计,而非在线干预。该框架将异构销售预测从启发式模型选择转变为可预测性模式和专家专业化的数据挖掘。代码可在https://github.com/hit636/FAME获取。

英文摘要:

Large-scale retail and industrial forecasting systems contain many heterogeneous time series whose lifecycle, sparsity, volatility, seasonality, spectral patterns, and contextual sensitivity differ substantially. A single forecasting model rarely performs well across all regimes, while dense ensembles increase inference cost and provide limited insight into expert suitability. This paper studies forecastability-aware expert routing: learning how data characteristics determine the suitability of forecasting experts. We propose \method{}, a sparse mixture-of-experts framework that represents each series with a multidimensional forecastability fingerprint, mines expert-suitability targets from validation performance, and trains a cost-aware sparse router to activate a small budgeted set of experts for each series. Using a production-scale vending-machine sales dataset from Shandong New Beiyang (SNBC), where the forecasting component has been integrated into the replenishment-planning pipeline, together with public retail benchmarks, we show that expert suitability varies systematically across data regimes. On the industrial dataset with 5,000+ machines and 60M+ transactions, \method{} Top-2 reduces MSE by 12.4\% over the strongest single expert, LightGBM, while executing 1.92 experts per series on average. The deployed component produces demand forecasts, while inventory-oriented gains are estimated by an offline replay simulator under a fixed replenishment policy rather than by online intervention. The framework turns heterogeneous sales forecasting from heuristic model selection into data mining of forecastability patterns and expert specialization. Code is available at https://github.com/hit636/FAME

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑