Neural techniques for ST-T change detection
暂无分享,去创建一个
The static neural approach (sNN), analyzing the signal beat by beat, and the dynamic neural approach, processing sequences of beats and implemented by means of Recurrent Neural Networks (RNN), are here compared on the basis of the European ST-T database. In order to perform the learning process, the high number of free parameters in the RNN is reduced by embedding "a priori" knowledge into the structure, and the input space dimension of the sNN is decreased by applying the PCA. The final results are comparable, even though the high number of free parameters in the recurrent approach produces a slow learning process and allows only to use simple data sets, far from being realistic.
[1] Barak A. Pearlmutter. Learning State Space Trajectories in Recurrent Neural Networks , 1989, Neural Computation.
[2] G. Moody,et al. The European ST-T database: standard for evaluating systems for the analysis of ST-T changes in ambulatory electrocardiography. , 1992, European heart journal.
[3] G.B. Moody,et al. Analysis of the cardiac repolarization period using the KL transform: applications on the ST-T database , 1994, Computers in Cardiology 1994.