This is an implementation of a neural network with hidden units that have sigmoid functions as their output. The ouput does not have sigmoid function because it is a model that is predicting a continuious variable. More details in the Jupyter Notebook.
The Neural Network has calculations that do both Forward Propagation through to the output node and then do backwards propagation to calculate hidden units errors and adjust weights accordingly.
[Problem Statement]
This nueral Network model predicts the daily bike rental ridership in SFO Bay Area.
Link to Project
[Problem Statement] This model uses tensorflow to identify handwritten images in the MNIST dataset
[Problem Statement] This model uses Keras and CNN to classify dogs and cats using the Kaggle dataset here
We also use the Resnet10 pretrained model that is part of the Keras library to identify what kind of dog or cat it is.
[Problem Statement] This model uses Keras and CNN to classify clothing images Fashion MNIST dataset here