Application of BP neural networks to transition detection in time series models
Takahiro Emoto, Masatake Akutagawa, Yohsuke Kinouchi, Udantha Ranjith Abeyratne, Hirofumi Nagashino · 2005
Biological signals are used in medical field to assess and track the functional states of vital organs such as the brain. The complexity (eg. nonlinearity and non-stationarity) of such signals and their low signal to noise ratios often make it a challenging task to use them in time critical applications. In this paper we propose a new neural network based technique to address those problems. We show mat a feedforward, multilayered neural network can conveniently capture the states of a nonlinear systems in its connection weight-space, after a process of supervised training. The performance of the proposed method is investigated with some systems simulated via a mathematical model, and the system generating real world Electroencephalogram (EEG) signals.