Task: Implement and Demonstrate Inverted Dropout Scaling

Task: Implement and Demonstrate Inverted Dropout Scaling: a task in Terminal-Lego-15k (Harbor dataset). In neural network training, dropout is a regularization technique that randomly sets neuron outputs to zero with some probability. There are two common approaches: