Research on QoE prediction model based on automatic machine learning
Accurately predicting the users"quality of experience(QoE)for streaming video is crucial for enhancing its service.QoE prediction models for streaming media typically rely on objective metrics such as video quality and network transmission quality.However,the subjectivity of users"QoE poses significant challenges for accurate assessment.In order to more precisely predict user experience quality,this paper introduces,for the first time,the application of automated machine learning to QoE prediction for streaming video,proposing an automated machine learning-based QoE prediction model.The model utilizes feature analysis to select optimal features from video quality assessment metrics and network quality assessment metrics as input,employing the H2O AutoML automated machine learning algorithm for QoE modeling.To evaluate the effectiveness of the method,the experiments are conducted on the publicly available SQoE-Ⅲ database,comparing the results with a traditional machine learning XGBoost-based QoE model.The experimental results demonstrate the significant progress in QoE prediction by adopting the automated machine learning-based model through automatic feature selection and model tuning.The model's MAE is 5.53699,and RMSE is 7.35987,effectively improving the accuracy of QoE prediction.Therefore,this study provides new perspectives and methods for QoE modeling,contributing to a deeper understanding and precise prediction of user perceptual satisfaction with streaming video.
streaming videoquality of experienceautomatic machine learningmachine learning
NETL
NSTL
万方数据
