Boosting feed-forward neural network for internet traffic prediction
Hanghang Tong, Chongrong Li, Jing-rui He · 2005
Internet traffic prediction plays a fundamental role in network design, management, control, and optimization. The self-similar and non-linear nature of network traffic makes high accurate prediction difficult in this paper, boosting is introduced into traffic prediction by considering it as a classical regression problem. A new scheme together with its adaptive version is proposed to update weight distribution. The new scheme controls the update rate by a parameter, while its adaptive version introduces no extra parameter and is adaptive to the training error of basic regressors and the current iteration number. Experimental results on real network traffic which exhibits both self-similarity and non-linearity demonstrate the effectiveness of our method.