Yet another demo for begining to work with RNNs....
A simple implentation of a Linear Layer in pyTorch is available in simpleMLP.py.
In the other file, a sine waye is generated and a RNN model is trained to detect the zero crossing zones. Although detecting zero cross can be easily and effieciently handled with hard computing techniques, it should be noted that more complex patterns can't be detected easily. Also, this Repo aims to demostrate the implementation of RNNs in PyTorch.
Train signal and the output are show below:
Train sine wave is generated at 1.8 Hz and is sampled at rate of 588 samples/second.
The same model creates output on a test signal as below:
