Title :
Adaptive trained neural networks for traffic prediction of VBR MPEG-2 video sources
Author :
Doulamis, Nikolaos D. ; Doulamis, Anastasios D. ; Kollias, Stefanos D.
Author_Institution :
National Technical University of Athens, Dept. of Electrical and Computer Engineering, 9, Heroon Polytechniou str., Athens, Greece
Abstract :
In this paper, a novel adaptively trained neural network structure is proposed for traffic rate prediction in MPEG-2 video sequences. The algorithm automatically adapts the network weights in case of highly rate changes taking into account both the current and the previous traffic samples. Application of the proposed scheme to real life sequences has shown a superior performance compared to other traditional methods.
Keywords :
Adaptive systems; Bit rate; Neural networks; Predictive models; Streaming media; Training; Transform coding;
Conference_Titel :
Signal Processing Conference, 2000 10th European
Conference_Location :
Tampere, Finland
Print_ISBN :
978-952-1504-43-3
