基于注意力与神经网络的视频流行度预测模型-学报杂志社

期刊信息

合肥工业大学（自然科学版），2023年11月，第46卷第11期：1472-1478

DOI: 10.3969/j.issn.1003-5060.2023.11.006

作者信息

马学森 $ ^{1,2} $，杨智捷 $ ^{1,2} $，储昭坤 $ ^{1,2} $，周天保 $ ^{1,2} $

(1. 合肥工业大学计算机与信息学院，安徽合肥 230601；2. 安全关键工业测控技术教育部工程研究中心，安徽合肥 230601)

摘要和关键词

摘要: 针对传统预测算法预测精度低及难以处理多变量的时序数据等缺点,文章提出一种采用双向长短期记忆(bi-directional long short-term memory, BiLSTM)网络和时间模式注意力(temporal pattern attention, TPA)机制相结合的视频流行度预测模型。双向长短期记忆网络从视频流行度时间序列的正向和反向提取时间特征,时间模式注意力机制从双向长短期记忆网络输出状态的深层特征提取时间模式,有利于视频流行度预测。真实视频数据的实验结果表明,与经典时序预测方法相比,TPA-BiLSTM模型能够有效地降低预测的误差,提高预测的准确性。

关键词: 流行度预测；多元时间序列；双向长短期记忆(BiLSTM)网络；注意力机制；卷积神经网络(CNN)

Authors

MA Xuesen $ ^{1,2} $, YANG Zhijie $ ^{1,2} $, CHU Zhaokun $ ^{1,2} $, ZHOU Tianbao $ ^{1,2} $

(1. School of Computer Science and Information Engineering, Hefei University of Technology, Hefei 230601, China; 2. Engineering Research Center of Safety Critical Industrial Measurement and Control Technology of Ministry of Education, Hefei 230601, China)

Abstract and Keywords

Abstract: In view of the shortcomings of traditional prediction algorithms such as low prediction accuracy and difficulty in processing multivariate time series data, a video popularity prediction model using a combination of bi-directional long short-term memory (BiLSTM) and temporal pattern attention (TPA) is proposed. BiLSTM extracts temporal features from the forward and reverse of the video popularity time series, while TPA extracts temporal patterns from the deeper features of the output states of BiLSTM, which is beneficial for video popularity prediction. Experimental results on real video data show that the proposed TPA-BiLSTM combined model effectively reduces the prediction error and improves the prediction accuracy compared with classical time series prediction methods.

Keywords: popularity prediction; multivariate time series; bi-directional long short-term memory(BiLSTM); attention mechanism; convolutional neural network(CNN)

基金信息

基金项目：国家重点研发计划资助项目（2020YFC1512601）；广东省科技发展专项基金资助项目（2017A0101001）；安徽省自然科学基金联合基金资助项目（2008085UD08）和中央高校基本科研业务费专项资金资助项目（PA2021GDGP0061）

合肥工业大学学报(自然科学版)

导航菜单

基于注意力与神经网络的视频流行度预测模型