Login to unlock
Deep Learning · Phase 1: Neural Network Foundations · From Linear Models to Neurons

From Linear Models to Neurons — Checkpoint Quiz

Checkpoint quiz 3 questions
Back to course

Check your understanding of this module before moving on.

?

Checkpoint quiz 3 questions

A quick check on this module's key ideas before you move on
0/3
Question 01

What is the key difference between a perceptron and a single artificial neuron used in modern deep learning?

B is correct. The perceptron's step function has a derivative of zero almost everywhere, so gradients can't flow through it. Replacing the step with a smooth activation is what lets backpropagation work.
Question 02

Why does a single artificial neuron, by itself, still only compute a linear decision boundary?

A is correct. A neuron computes w·x + b and then applies an activation. Whether the overall boundary is linear or not depends entirely on that activation; stacking many such neurons in layers is what lets a network represent curved boundaries.
Question 03

What does the bias term in a neuron let the model do that weights alone cannot?

A is correct. Without a bias, every neuron's pre-activation value is forced to equal zero whenever all inputs are zero. The bias is a free offset that lets the boundary sit anywhere, not just through the origin.