计算机科学
电流(流体)
深度学习
比例(比率)
系列(地层学)
数据集
数据挖掘
人工智能
时间序列
特征(语言学)
空间相关性
数据建模
集合(抽象数据类型)
机器学习
地质学
地理
程序设计语言
哲学
古生物学
海洋学
数据库
电信
地图学
语言学
标识
DOI:10.1109/cisai54367.2021.00101
摘要
Realizing the accurate prediction of ocean current has very important scientific value, practical application value and research significance. Due to the large scale and complexity of ocean data sets, traditional methods for ocean current prediction have many challenges in terms of efficiency and accuracy. It is proved that the deep learning method can learn the spatiotemporal characteristics of a large amount of data, and it is very effective and accurate to predict data with complex structures. In this paper, we present a new model structure STAGRU, which includes Spatial Module to extract spatial characteristic and the Attention Model to extract the nearest neighbor time correlation information based on the GRU. We conduct a series of experiments on the offshore data set of eastern China, and compare with the previous models. The results show that our model outperforms many baselines in RMSE on the component of U and V has been increased by 5.85% and 3.40% respectively.
科研通智能强力驱动
Strongly Powered by AbleSci AI