The Perceptron
A perceptron is one artificial neuron. It takes some numbers, mixes them, and says yes or no. Every deep network in this track is built from many of these tiny parts. In this lesson you work one by hand, watch it learn, and see the one thing it cannot do.
New words
- Perceptron
- One artificial neuron that turns some inputs into a yes or no.
- Weight
- A number that says how much one input matters.
- Bias
- A number added at the end. It moves the point where the answer flips.
- Step function
- A rule that gives 1 if the number is 0 or more, and 0 if not.
- Decision line
- The line that splits the points the neuron calls 1 from those it calls 0.
- Learning rule
- A small update to the weights each time the neuron gets a point wrong.
- XOR
- A yes or no puzzle that is 1 when the two inputs differ.
What you will learn
- Should I carry an umbrella?
- Inputs, weights and a bias
- The step function gives 0 or 1
- The line a perceptron draws
- How does a perceptron learn?
- Try it: watch the line move
- What a perceptron cannot do: XOR
- Common mistakes
- Explain it to a friend
- Key takeaways
10 quiz questions at the end.
The full lesson, with interactive demos and the quiz, opens after you sign in with Google. It is free.