Exercises
These exercises check the activation and nonlinearity ideas from the chapter.
DL-C04-C-001Exercise: ReLU vector
Apply ReLU to [-3, 2, 0, 5]. How many outputs are positive?
Compute it first, then check your number.
DL-C04-C-002Exercise: ReLU value
What is relu(-11)?
Compute it first, then check your number.
DL-C04-C-003Exercise: Sigmoid range
Enter 1 if sigmoid outputs are bounded between 0 and 1, or 0 if they are unbounded.
Compute it first, then check your number.
DL-C04-C-004Exercise: Detector
Let x = [2, 2], w = [1, -3], and b = 5. Compute relu(x . w + b).
Compute it first, then check your number.
DL-C04-C-005Exercise: Linear collapse
If h = 4x and y = -2h, then y equals what multiple of x?
Compute it first, then check your number.
DL-C04-C-006Exercise: Why nonlinearity matters
Enter 1 if nonlinear activations help depth increase the model family, or 0 if stacked linear layers alone are enough.
Compute it first, then check your number.