The purpose of this code is to train a neural network to label a cosine waves that are at fixed frequencies but different phases and are corrupted with noise.
The classifier is built using a deep neural network with five layers with cross entropy as the loss function.
The noise consists of two components, gaussian white noise and harmonic noise. The harmonic noise selects a random number (up to 12) of cosine waves at varying frequencies and are added to the original cosine wave of interest.
To run the neural network, run the dnn_denoiser.py file.
The model reports back about 96-97% accuracy when sampled at 112 samples per unit of time with four different phases.
TensorFlow v1.0
Numpy v1.11
This application uses Open Source components. You can find the source code of their open source project along with licence information below. We acknowledge and are grateful to these developers for their contribution to open source software.
Project: tensorflow-mnist-tutorial (https://github.com/martin-gorner/tensorflow-mnist-tutorial) Specifically (https://github.com/martin-gorner/tensorflow-mnist-tutorial/blob/master/mnist_2.2_five_layers_relu_lrdecay_dropout.py) Licence: Apache Licence 2.0