Prediction method of user experience quality based on multi-layer neural network in video service

Abstract

The invention discloses a method for predicting user experience quality based on a multi-layer neural network in video services, including data preprocessing, that is, selecting characteristic parameters affecting user experience in video services, and mapping them into User's QoE. Then the QoE prediction model of multi-layer neural network is established. The neural network consists of five layers, which are in order from low to high: input layer-first hidden layer-second hidden layer-third hidden layer-output layer. Input the preprocessed data, obtain the optimal parameter value of the model, and train the neural network model established above. Finally, the user experience quality QoE prediction is completed. The present invention preprocesses the data, selects important feature attributes, and comprehensively considers various parameters, so that the video quality predicted by the model can be really close to the user's subjective experience of video quality, which helps to better predict the user's experience quality and facilitates timely Accurate feedback results help service providers and network operators to continuously improve video services and transmission services.

Description

technical field [0001] The invention relates to user experience quality prediction, in particular to a user experience quality prediction method based on a multi-layer neural network in video services. Background technique [0002] The rapid development of Internet technology enables people to access various multimedia services, and in particular, IPTV now provides various services, making people's lives colorful. But on the other hand, service providers and network operators are more concerned about the quality of the provided video services, that is, what is the IPTV user experience of watching videos. This makes the prediction and evaluation of user experience quality a hot spot that service providers and network operators pay attention to. Quality of Experience (QoE) is defined as "the overall acceptability of an application or service as perceived by the end user". QoE is not only affected by the service itself, but also by the user's environment. Because machine lea...

AI Technical Summary

Problems solved by technology

However, traditional machine learning methods, such as support vector machines and decision trees, have very limited predictive performance and cannot accurately complete this task. Therefore, it is necessary to design new models and predictive methods to complete the prediction and improvement of IPTV user experience quality

Images