Performance of perceptron predictors for lossless EEG signal compression

Abstract

In this paper, the performance of different types of perceptron predictors for EEG signal compression is investigated. A two-stage lossless compression scheme which involves the decorrelation of EEG samples in the first stage and entropy coding in the second stage is considered. The second stage employs an arithmetic coding scheme. A comparison of the performance of the perceptron predictors with that of linear predictors such as FIR, NLMS is presented. It is found that the single-layer perceptron performs, in general, better than the multi-layer perceptrons as well as linear predictors.