Description: From scratch implementation of a single hidden layer neural network and perception for fitting continuous function
Dataset: A somewhat continuous, somewhat classical regression dataset with a single input and output value
Neural Network: Neural network contains a single hidden layer. Number of hidden units can be adjusted through parameters. Activation function is a sigmoid, and model is trained with batch gradient descent.
Percetron: Single input single output perceptron with an identity activation funtion
Loss function: MSE
Repo contains both a notebook and .py files