In numerous applications, for instance in predictive maintenance, there is a
pression to predict events ahead of time with as much accuracy as possible
while not delaying the decision unduly. This translates in the opti
本文提出了基于 End-to-End Learned Early Classification of Time Series (ELECTS) 的模型,通过训练远程感知卫星获取的时间序列数据来进行预测并提高农作物分类的准确性和判断速度。该模型不仅可以提高分类准确性,同时还能在不降低准确性的情况下尽可能地减少需要下载、存储和处理的数据量。
提出了一种新的时间编码和模型结构,分别是 Time Encoding 机制和 Time Encoding Echo State Network (TE-ESN),该结构能够处理不规则时间序列,同时可以将长短期记忆和串联处理融入到模型中,用于更准确的预测。在一个混沌系统和三个实际数据集的实验证明,TE-ESN 比所有基线的表现更好,并具有更好的遗传特性。
Ubiquitous sensors produce high frequency streams of numerical measurements which can be segmented using the ClaSS algorithm based on self-supervised time series classification and statistical tests to detect change points.