融合统计特征与深度编码，实现时间序列分类的持续学习。
Combining Statistical Features and Deep Encodings for Rehearsal-Based Class-Incremental Time Series Classification
- 双流特征提取：用预训练模型生成时序嵌入，同时结合统计特征。
- 平均准确率达标，遗忘率低，五数据集表现稳定。
- 适合需要持续更新类别的时间序列应用，如工业监控、医疗诊断。
真实场景中的许多系统需要在不遗忘已有知识的前提下，不断添加新类别和新信息。这被称为类别增量持续学习，而多变量时间序列由于其时间结构的复杂性，使该问题更加困难。本文提出一种新方法，通过构建双流特征提取管道（利用预训练冻结基础模型生成的深层时序嵌入特征，以及统计特征），实现多变量时间序列的类别增量持续学习。在五个基准数据集上的实验表明，所提系统在所有数据集上均达到具有竞争力的平均准确率，且在所有实验配置下保持较低的遗忘率。
原文摘要 · Abstract (English)
Many systems used in real-world environments require adding new categories and incorporating new information without forgetting what was previously learnt by the classification model. This is known as class-incremental continual learning, and in the case of multivariate time-series, is further complicated by the temporal structure of the data. In this paper, we present a novel approach for performing class incremental continual learning for the classification of multivariate time series data based upon the construction of a dual-stream feature extraction pipeline (using both deep temporal embedding features generated via a pre-trained frozen foundation model and application of statistical features). Evaluated on five benchmark datasets, the proposed system achieves competitive average accuracy across all datasets while maintaining low forgetting rates across all experimental configurations.
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。