AI & Machine Learning Backpropagation & Gradient Descent Quiz

This hard quiz covers how neural networks learn. You will apply the chain rule, compute gradients and update steps with small numbers, and answer the interview questions about learning rates, initialisation and vanishing gradients.

Hard 10 questions ~8 minutes Pass at 80%

Choose a mode

Playing as a guest: sign in to keep your XP and streak.

Sample question

Question 1 of 10

What does backpropagation compute?

Show the answer

D. The gradient of the loss with respect to every weight, using the chain rule from the output layer back Backprop reuses the values saved in the forward pass and applies the chain rule layer by layer, getting all gradients in about the cost of one more pass. The optimizer (such as SGD) then uses those gradients to update the weights.

What this quiz covers

  • Backpropagation as the chain rule from the loss backwards
  • The update rule w = w - lr * gradient
  • Learning rates that converge, oscillate or diverge
  • Batch vs mini-batch SGD and momentum
  • Symmetry breaking and the sigmoid's small derivative

Questions people ask

Do I need an account?

No. Anyone can play. Sign in to keep your XP, streak, badges and history.

Can I retake it?

Yes. Your best score is kept, and the questions and answers are shuffled each time.