计算机工程与应用 ›› 2020, Vol. 56 ›› Issue (12): 237-242.DOI: 10.3778/j.issn.1002-8331.1903-0218

• 工程与应用 • 上一篇    下一篇

面向家居用户行为预测的BiGRU-DAtt模型研究

徐雅芸,曾碧,梁天恺,廖文雄   

  1. 广东工业大学 计算机学院,广州 510006
  • 出版日期:2020-06-15 发布日期:2020-06-09

Research on BiGRU-DAtt Model for Home User Behavior Prediction

XU Yayun, ZENG Bi, LIANG Tiankai, LIAO Wenxiong   

  1. School of Computers, Guangdong University of Technology, Guangzhou 510006, China
  • Online:2020-06-15 Published:2020-06-09

摘要:

针对目前智能家居用户行为预测方法准确率低、通用性差以及缺乏人性化的问题,提出一种基于BiGRU-DAtt模型的智能家居用户行为预测方法。该方法根据智能家居用户操控行为数据具有服从幂律分布与对称性两大特点,基于双向门控循环(BiGRU)神经网络挖掘用户操控行为之间的关系,基于注意力机制重点关注一定范围内具有对称性的操控行为,使用真实用户操控记录进行对比实验。结果表明该方法能够充分挖掘用户操控智能设备之间的关联关系以及用户的行为习惯,实现高准确率的用户行为预测。

关键词: 智能家居, 双向门控循环神经网络, 注意力机制, 行为预测

Abstract:

Aiming on the home user behavior prediction method that currently remains the problem of low accuracy, poor versatility and for lacking of humanization sense, this paper proposes a smart home user behavior prediction method based on BiGRU-DAtt model, which has two characteristics of power law distribution and symmetry according to the smart home user manipulation behavior data. This method adopts the Bidirectional Gated Recurrent Unit(BiGRU) neural network to figure out the relationship between user manipulation behaviors. The symmetry control behavior is focused within a certain range by using the attention mechanism. The comparison experiment is conducted with the real user control data. The experimental results show that this method can fully explore the association between user-controlled intelligent devices and user’s behavior habits, and achieve high accuracy user behavior prediction.

Key words: smart home, Bidirectional Gated Recurrent Unity(BiGRU), attention mechanism, behavior prediction