TensorFlow Machine Learning Projects

Deep neural network module

The DNN module is implemented with the Convolutional Neural Network (CNN) architecture. The code implements multiple variations of ConvNet, each variation producing different levels of accuracy and taking a different amount of time to train.

The code for building the model is provided in the models.py file. It allows the creation of four different models, depending on the parameter passed at the command line:

• single_fc: This model has only one fully connected layer.
• conv: This model is a full CNN architecture with two pairs of Convolution and MaxPool layers, followed by a fully connected layer.
• low_latency_conv: This model has one convolutional layer, followed by three fully connected layers. As the name suggests, ...