Clever Geek Handbook

Wavelet Neural Networks: Why Is There No Inverse Transformation?

I am wondering why there is no inverse transform in the Wavelet Neural Network that recomposes the signal?

Why are only wavelet coefficients enough to find the desired signal?

Wouldn't it be better to have an IWT?

Jeff

+3


source to share


2 answers


Using a wavelet in a neural network is just for feature extraction. And from the decomposition of the wavelets, you can reconstruct the original signal. One more question about your question. Are you using WNN for regression or classification? In the case of classification, there is no need for an IWT. In regression case, I have no idea how to set WNN.



proposal is acceptable: Wavelet neural networks

0


source


I have a question that has confused me for a long time, many people have recommended the link " http://maths.york.ac.uk/www/sites/default/files/Veitch.pdf ", Neural Network Wavelet. But when I open it on China net, it shows the link " https://www.york.ac.uk/maths/ " which is the home page for the Department of Mathematics at York University. Is there something wrong with this link or is there a network limitation issue? Can anyone send me the documents? Thanks a lot , my mailbox is amberjinb628@163.com.



0


source






More articles:


All Articles