基于迁移学习的电视观众分类模型  

TV Audience Classification Model Based on Transfer Learning

作　　者：卢川琴 王国永 张立山 LU Chuanqin;WANG Guoyong;ZHANG Lishan(Shandong provincial Radio and Television Monitoring Center,Jinan 250014,China)

机构地区：[1]山东省广播电视监测中心,山东济南250014

出　　处：《电视技术》2024年第9期31-34,共4页Video Engineering

摘　　要：针对智能电视机因无法获取电视观众特征数据导致无法准确地对观众进行分类的问题,提出一个基于迁移学习的电视观众分类模型。先分析用户特征和其喜好的电视节目,并训练基模型,再使用真实的电视观众数据和其喜爱观看的电视节目数据,将基类型迁移到电视观众分类任务中。实验证明,所提模型具有良好的分类准确性。In response to the problem that smart TV cannot obtain TV audience feature data,which makes it difficult to accurately classify viewers,this article proposes a TV audience classification model based on transfer learning.Firstly,analyze the characteristics of users and their preferred TV programs,and train a base model,then,using real TV audience data and their favorite TV programs,transfer the base type to the TV audience classification task.Experimental results have shown that the proposed model has good classificationaccuracy.

关 键 词：电视观众分类 迁移学习 深度神经网络 

分 类 号：TP311.1[自动化与计算机技术—计算机软件与理论]