COMMON OPTIMIZATION OF ADAPTIVE PREPROCESSING UNITS AND A NEURAL-NETWORK DURING THE LEARNING PERIOD - APPLICATION IN EEG PATTERN-RECOGNITION

In this study, a proposition of simultaneous training of the neural ne twork (multilayer perceptron) and adaptive preprocessing unit is prese nted. This cooperation enables the network to affect the preprocessing and as a consequence to vary the locations of pattern vectors in a fe ature space. Thus, during the learning process the network tries to fi nd a good separation of classes of patterns, which results in converge nce of the whole learning process. The strategy was developed in order to make efficient EEG monitoring in neonates possible. A comparison o f the method presented herein with the known learning strategies for n eural networks shows the need for using it as an alternative learning process. The convergence of the whole system is also discussed. (C) 19 97 Elsevier Science Ltd.