(英) |
In recent years, machine learning such as deep learning has attracted attention. Various applications have been proposed and studied in the wireless communication field. Therefore, in the objective communication quality estimation, we consider a method that uses evaluation parameters specific to wireless communication such as propagation error, interference, transmission, throughput, packet loss, etc. for machine learning. In this paper, we focus especially on weight adjustment of neural networks. Conventionally, error back-propagation method etc. are used to adjust this weight, but the performance when using stochastic optimization method is confirmed. Success-History based Adaptive DE (SHADE) was used as the optimization method. The purpose is to control the QoE index so that it always obtains a high value by using the QoE objective evaluation index of video distribution as the evaluation index. |