Detection of seizure activity in EEG by an artificial neural network: a preliminary study
Computers and Biomedical Research
IEEE Transactions on Pattern Analysis and Machine Intelligence
Biologically Inspired Architecture of Feedforward Networks for Signal Classification
Proceedings of the Joint IAPR International Workshops on Advances in Pattern Recognition
Hi-index | 0.00 |
Architecture of a neural network combining automatic feature extraction with the minimized amount of network training acquired by means of employing of a multistage training procedure is investigated. The network selects prototypical signals and calculates features based on the similarity of a signal to prototypes. The similarity is measured by the prognosis error of the linear regression model. The network is applied for the meaningful paroxysmal activity vs. background classification task anp provides better accuracy than the methods using manually selected features. Performance of several modifications of the new architecture is being evaluated.