Python implementation of neural network.
numpy
argparse
python neural_network.py
By default, this will load breast cancer dataset and train a neural network for classification. Model prediction score will be displayed.
Some options can be specified for the program:
-dataset
str, Dataset to be used, breast_cancer
or energy_efficiency
. Default: breast_cancer
-num_of_layers
int, Number of hidden layers in the network. Default: 2
-num_of_units
int, Number of neuron unit within each hidden layer. Default: 4
-lr
float, Learning rate. Default: 1e-3
-tolerant
float, Threshold for loss change when early stopping. Default: 5e-5
-max_it
int, Maximum iteration time. Default: 500
Example:
python neural_network.py -dataset energy_efficiency -num_of_layers 5