arXiv cs.AI· Julien Guit\'e-Vinet, Alexandre Blondin Mass\'e, \'Eric Beaudry·· 11 小时前AI 评分22
一种轻量级多视图短期负荷预测方法
A Lightweight Multi-View Approach to Short-Term Load Forecasting
AI 导读
研究者提出一种轻量级多视图短期负荷预测方法，利用单值嵌入和缩放时间范围输入高效捕捉时间相关特征，并引入嵌入向量 dropout 机制防止模型过度依赖特定特征、提升可解释性。该方法以显著更少的参数取得有竞争力的性能，在多个数据集（含噪声或稀疏数据场景）上表现稳健，并能揭示各特征对预测的贡献。
正文 · 原文
Abstract:Time series forecasting is a critical task across domains such as energy, finance, and meteorology, where accurate predictions enable informed decision-making. While transformer-based and large-parameter models have recently achieved state-of-the-art results, their complexity can lead to overfitting and unstable forecasts, especially when older data points become less relevant. In this paper, we propose a lightweight multi-view approach to short-term load forecasting that leverages single-value embeddings and a scaled time-range input to capture temporally relevant features efficiently. We introduce an embedding dropout mechanism to prevent over-reliance on specific features and enhance interpretability. Our method achieves competitive performance with significantly fewer parameters, demonstrating robustness across multiple datasets, including scenarios with noisy or sparse data, and provides insights into the contributions of individual features to the forecast.
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2602.09220 [cs.LG] |
| (or arXiv:2602.09220v2 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2602.09220 arXiv-issued DOI via DataCite |
Submission history
From: Julien Guité-Vinet [view email]
[v1]
Mon, 9 Feb 2026 21:43:42 UTC (1,804 KB)
[v2]
Tue, 3 Nov 2026 23:30:20 UTC (5,501 KB)
来源：arXiv cs.AI · arxiv.org