用符号图结构提升多变量时间序列预测的模型表现
Multivariate de Bruijn Graphs: A Symbolic Graph Framework for Time Series Forecasting
- 将连续序列转为符号图，用 de Bruijn 图建模时间依赖
- 在真实数据集上实现比基线模型更低的 MAE 与 RMSE
- 适合需要结构感知的时序预测任务，如金融、气象
时间序列预测对基础模型而言仍具挑战，源于时间异质性、高维性和缺乏固有符号结构。本文提出 DRAGON（基于 de Bruijn 图的离散表示与增强图编码），引入多变量 de Bruijn 图（MdBGs）连接符号表示与神经建模。该方法将连续输入序列离散化并映射到固定图结构上，通过图注意力机制实现动态上下文恢复。作为双分支架构中的辅助模块，DRAGON 为传统 CNN 编码器补充了符号化、结构感知表示。所有代码已开源：https://github.com/KurbanIntelligenceLab/MultdBG-Time-Series-Library
原文摘要 · Abstract (English)
Time series forecasting remains a challenging task for foundation models due to temporal heterogeneity, high dimensionality, and the lack of inherent symbolic structure. In this work, we propose DRAGON (Discrete Representation and Augmented Graph encoding Over de BruijN Graphs), a novel encoder that introduces Multivariate de Bruijn Graphs (MdBGs) to bridge the gap between symbolic representations and neural modeling. DRAGON discretizes continuous input sequences and maps them onto a fixed graph structure, enabling dynamic context recovery via graph-based attention. Integrated as an auxiliary module within a dual-branch architecture, DRAGON augments conventional CNN-based encoders with symbolic, structure-aware representations. All code developed for this study is available at: https://github.com/KurbanIntelligenceLab/MultdBG-Time-Series-Library
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。