22 - Quiz¶

• Your name:

• Your VT email:

What do the parameters \(\theta\) correspond to in a neural network \(f(x; \theta)\)?

• A. Only the input data
• B. The collection of weights and biases
• C. Only activation functions
• D. The loss function

What is the purpose of the forward method?

• A. To update weights
• B. To compute gradients directly
• C. To initialize model parameters
• D. To define the computation \(f(x; \theta)\)

Why are activation functions necessary?

• A. They reduce memory usage
• B. They make the model linear
• C. They introduce nonlinearity into the model
• D. They reduce redundant parameters in the model

What does the optimizer of a neural network do?

• A. Computes the loss
• B. Loads training data
• C. Computes gradients to update model parameters
• D. Updates model parameters using gradients