Results (
Thai) 1:
[Copy]Copied!
C. Signal Classification with Perceptron A problem of particular interest to electrical engineers is that of signal detection, particularly in a noisy environment. Methods such as filtering and signal averaging have been used successfully. In this example a neural net is first trained to learn the input patterns of three signals - a square wave, a triangular wave, and a sine wave. Then, noise is superimposed on these signals with the aid of MATLAB's random number generating function. Each signal is then tested to see whether the network can properly classify it when a certain level of noise is added to it. This is analogous to training a net to learn alphabetic characters and then trying to classify characters with "errors", as discussed previously.Figure 5 shows samples of the plotted waveforms for the case where the signal to noise amplitude ratio is 5.
Being translated, please wait..