计算机科学
变压器
时间序列
计算
系列(地层学)
人工智能
数据挖掘
算法
机器学习
工程类
电气工程
生物
古生物学
电压
作者
Lin Yang,Irena Koprinska,Mashud Rana
标识
DOI:10.1007/978-3-030-63836-8_51
摘要
In this paper, we present SpringNet, a novel deep learning approach for time series forecasting, and demonstrate its performance in a case study for solar power forecasting. SpringNet is based on the Transformer architecture but uses a Spring DTW attention layer to consider the local context of the time series data. Firstly, it captures the local shape of the time series with Spring DTW attention layers, dealing with data fluctuations. Secondly, it uses a batch version of the Spring DTW algorithm for efficient computation on GPU, to facilitate applications to big time series data. We comprehensively evaluate the performance of SpringNet on two large solar power data sets, showing that SpringNet is an effective method, outperforming the state-of-the-art DeepAR and LogSparse Transformer methods.
科研通智能强力驱动
Strongly Powered by AbleSci AI