4.2.4. Model Training and Optimization

Try these first

1. 1.

   What is gradient descent?

   Hint

   Think about how the model learns from its mistakes.

2. 2.

   Define overfitting.

   Hint

   Consider what happens if a student memorizes answers without understanding.

3. 3.

   What is the purpose of gradient descent?

   • To increase error
   • To minimize the loss function
   • To select hyperparameters

   Hint

   Think of this as finding the lowest point in a valley.

4. 4.

   True or False: Backpropagation is an algorithm used to clean data before training.

   • True
   • False

   Hint

   Recall the role of backpropagation in training.

5. 5.

   Given a dataset with high dimensions, outline a strategy to reduce the risk of overfitting when training a model.

   Hint

   Consider ways to simplify the model.

6. 6.

   Suppose two models are trained with different learning rates. One has a high learning rate, and the other has a low learning rate. Predict the outcomes and justify your reasoning regarding convergence and model performance.

   Hint

   Think about how quickly or slowly adjustments are made during training.