Feedforward Neural Network (MLP)
1. How many input features does the Iris dataset have in this experiment?
2. Which of the following best describes the Iris dataset class distribution?
3. What does stratify ensure when splitting a classification dataset?
4. Which technique checks the correctness of backpropagation gradients numerically?
5. How many hidden layers does the MLP architecture contain?
6. Which loss function is appropriate for the compiled MLP on one-hot labels?
7. What activation is typically used in the output layer for this 3-class problem?
8. Which technique reduces overfitting by adding a penalty proportional to the squared weights?
9. Which optimisers are compared for finding the best accuracy in this experiment?
10. What does forward propagation produce before the computation of loss?