theraizer / binary-artificial-neural-network

ANN created in python. Easy training and testing through the console.

Artificial Neural Network (Binary)

Summary

Uses console input to train and test an ANN on an iris and heart disease data set.

Accuracy

• 99.99999999999997% accuracy on iris data set.
• 88.88888888888889%% accuracy on heart disease data set.

Layers

• Uses ReLu activations for hidden layers and Sigmoid for the final layer.

Hyper Parameters

• alpha (learning rate)
• # of iterations
• lambd (controlling L2 regularization)
• batch_count (the size of a mini batch)
• decay_rate (alpha decay rate)
• beta_1 (momentum)
• beta_2 (RMS)