Reference pattern weight initialization for equalization

Abstract

The problem of weight initialization in multilayer perceptron networks is considered. A computationally simple weight initialization method based on the usage of reference patterns is investigated in channel equalization application. On one hand, the proposed method aims to set the initial weight values to be such that inputs to network nodes are within the active region. On ] the other hand, the goal is to distribute the discriminant functions formed by the hidden units evenly into the input space area where training data is located. The proposed weight initialization is tested in the channel equalization application where several alternatives for obtaining suitable reference patterns are investigated. A comparison with the conventional random initialization shows that significant improvement in convergence can be achieved with the proposed method. In addition, the computational cost of the initialization was found to be negligible compared with the cost of training.