How Neural Networks Learn · Lesson 1 · 3 minutes

Scale with weights

Shared by Hulayra

Multiply an input by its weight

Play this lesson in your browser. No account needed.

  1. Scale with weights

    Tunde checks traffic photos in Lagos. His neural model uses weights to scale each input.

    Scale with weights. Tunde checks traffic photos in Lagos. His neural model uses weights to scale each input.
  2. Weights multiply inputs

    A weight scales an input by multiplying it. A larger weight makes an input stronger.

    Weights multiply inputs. A weight scales an input by multiplying it. A larger weight makes an input stronger.
  3. Boost the signal to reach 12.

    This one uses your voice or your writing. It is in the Breni app.

  4. An input has value 3. Its weight is 4. What is the output?

    • 12
    • 7

    Answer: 12

    Weights multiply the input strength, they do not add to it.

  5. What does a weight of -2 do to an input of 5?

    • It gives -10
    • It cuts the connection to 0

    Answer: It gives -10

    Negative weights reverse and scale the input signal.

  6. Show how a weight alters an input feature.

    Answer: Pixel count multiplied by weight scaled signal

    Weights scale inputs to measure feature importance.

  7. Complete the statement about feature importance in code.

    signal = feature ___ weight

    Answer: signal = feature * weight

    A zero weight removes the effect of an input completely.

  8. Trace how an input turns into a weighted output.

    This one uses your voice or your writing. It is in the Breni app.

What to remember

  • Weights scale inputs by multiplication.
  • A weight of zero turns off an input completely.

Next in this course: Shift with biases. It carries on in the Breni app.