Signal classifications using neural networks

Hello , can anyone help me with ideas about how to classify different time series signals using neural networks?

Are there any single input , single output classification network? For instance lets say I have some signals-

y1 = A1*sin(2*pi*f1*t);

y2 = A2*sin(2*pi*f2*t);

y3 = A3*sin(2*pi*f3*t)+A3*sin(2*pi*f4*t);

These signals are different in amplitude and frequencies. Can a neural network be designed that can classify these three type of signals in three different class?

Training -

Input Target

y1 ------- 1

y2 ------- 2

y3 ------- 3

Now I want to test some arbitrary signal and see if the network can classify it correctly. At least frequency wise.

Thanks in advance

[Greg Heath]

Are the amplitudes and frequencies known?

What is the dimensionality of the input vector?

An FFT approach might be more fruitful.

Answer 1

[mae]

i am having a similar issue .. i want to use neural networks for ECG signal classification and i am stuck

[Greg Heath]

Be specific:

What apriori information will you have?

Greg