Pruned Neural Network For Artifact Reduction In EEG Signal
T.V. Sreenivas, V.S. Unnikrishnan, D. Narayana Dutt · 2005
Artificial neural networks are promissing for separating signals from interference. However, their success depends on the amount of training data and network complexity which is prohibitively high. Pruned neural networks (PNN) are explored in this paper for the specific problem of separating artifacts from EEG signals. It is shown that PNN provides better performance than the fully interconnected network of similar complexity.