1
2
3

Build a Neuron

An artificial neuron is one building block in the net. Many of them make a neural net. The idea comes from real brain cells. Those cells are called biological neurons.

Step 1 of 3

What you'll learn in this level

  • What weights and bias do inside a single neuron
  • How a neuron turns several inputs into one number
  • Why the sigmoid function keeps the output between 0 and 1

How One Neuron Works

An artificial neuron is one building block in the net. Many of them make a neural net. The idea comes from real brain cells. Those cells are called biological neurons.

Every neuron gets several inputs. It does a calculation with them. At the end it gives out one output. The exciting part is the calculation in between.

The Parts of a Neuron

  1. Inputs

    These are the signals that come into the neuron. They come from other neurons. Or from the world outside.

  2. Weights

    Numbers that say how important each input is. A weight above zero speaks for the output. Experts call that excitatory. A weight below zero speaks against it. That is called inhibitory.

  3. Bias

    A number that is always added on. It works like a threshold. It helps the neuron decide better.

  4. Activation Function

    A rule for the very last step. It turns inputs, weights and bias into the output.

In this task you build your own artificial neuron. You will see exactly how it calculates.

Guess first, then experiment

You set one weight on a neuron to 0. What happens to that input?

How a Neuron Calculates Its Output:

  1. Multiply each input by its corresponding weight
  2. Add all these weighted inputs together
  3. Add the bias value
  4. Apply the activation function to get the final output