M-level ionic-channel signal processing using neural network

A neural network model based on the backpropagation training algorithm is used to process M-level steplike signals. The method is constructed in three main stages: the normalization stage, the neural network main body, and the output interpretation stage. Results on simulated examples are presented and are comparable to those of the traditional algorithms. The Walsh-Hadamard transformation used for input representation shows higher performance than the original data. It is also demonstrated that a single-layer network is capable of classifying a steplike signal.<>