Creating MLP neural network from scratch. Using iris dataset.

Created MLP neural network with input, one hidden, output layer.
Optimizer coded: mini-batch GD, Batch GD, Adam. Mini-batch and Batch can be used with momentum.
Activation function: tahn, sigmoid, relu. Tahn is computation expensive training is much slower then using sigmoid or relu. ReLu was showed to be much faster and optimizers can converge. Best model why hyper parameters:

• neurons hidden layer: 20
• epochs: 300
• learning rate: 0.01
• batch size: 64
• hidden layer activation function: relu
• output layer activation function: sigmoid
• weight distribution: Uniform
• momentum: true