基于深度学习的大数据时空序列挖掘方法
Research on Big Data Spatiotemporal Sequence Mining Method Based on Deep Learning
张峰1
作者信息
- 1. 江苏联合职业技术学院宿迁开放大学办学点,江苏宿迁 223800;江苏省宿迁中等专业学校,江苏宿迁 223800
- 折叠
摘要
针对时空序列挖掘任务中的诸多挑战,文章提出一种基于L2正则化的长短期记忆网络(Long Short-Term Memory,LSTM)方法,旨在增强传统LSTM模型在长短时序列中的建模能力.首先,分析了大数据时空序列挖掘方法的框架,包括数据输入、数据预处理、深度学习模型、正则化优化及数据挖掘结果.其次,介绍了 LSTM的基本原理和基于正则化的LSTM优化方法.最后,进行实验分析.
Abstract
In response to the many challenges in spatiotemporal sequence mining tasks,this article proposes a Long Short Term Memory(LSTM)method based on L2 regularization,aiming to enhance the modeling ability of traditional LSTM models in long short time sequences.Firstly,the framework of big data spatiotemporal sequence mining methods was analyzed,including data input,data preprocessing,deep learning models,regularization optimization,and data mining results.Secondly,the basic principles of LSTM and regularization based LSTM optimization methods were introduced.Finally,conduct experimental analysis.
关键词
深度学习/时空序列/挖掘Key words
deep learning/spatiotemporal sequence/excavate引用本文复制引用
出版年
2024
NETL
NSTL
万方数据