A neural network approach to the detection problem using joint time-frequency distributions
Donald B. Malkoff, Leon W. Cohen · International Conference on Acoustics, Speech, and Signal Processing · 2002
A neural network algorithm for the detection and classification of transients in noise is described. The inputs are time-frequency representations modified so their entry location into the network of nodes is independent of the starting time of the transient. The algorithm is three-layered and feedforward, and is capable of training or testing in a single pass.>