Machine Learning Basics

When the derivative dy/dx of y = (2x + 1)² at x = 1 is found by the chain rule with u = 2x + 1 and y = u², which value is correct?

  1. A6
  2. B9
  3. C12
Show answer and explanation

Answer: C. 12

By the chain rule, dy/dx = (dy/du) × (du/dx). dy/du = 2u and du/dx = 2, and at x = 1, u = 3, so dy/dx = 2 × 3 × 2 = 12. Backpropagation applies this chain rule step by step from the output side toward the input side to compute efficiently the gradient of the loss with respect to the weights of every layer. Each layer multiplies the gradient received from the layer after it by its own local derivative and passes the result to the layer before it.

  • A. This finds only dy/du = 2u = 6 and forgets to multiply by du/dx = 2.
  • B. This is the value of y at x = 1, 3² = 9, not the derivative.
  • C. Correct. 2u × 2 = 2 × 3 × 2 = 12.
Practice this topic Back to topic

Related calculators

Books on Machine Learning Basics Ad

As an Amazon Associate I earn from qualifying purchases.

More Machine Learning Basics questions

  1. In a neural network with many stacked layers, which of the following is the reason that using the sigmoid function as the activation function makes the layers near the input hard to train?
  2. A model is trained on 10,000 training samples with a mini-batch size (batch size) of 50. How many weight updates are there in one epoch?
  3. How many trainable parameters does a fully connected layer with 64 inputs and 32 outputs (one bias per output) have?
  4. A 28×28 single-channel image is convolved with a 5×5 filter (stride 1, no padding), followed by 2×2 max pooling (stride 2). What are the height and width of the output?
  5. Which of the following correctly describes recurrent neural networks (RNNs) and LSTMs?