How a Neuron Calculates Output

The Complete Calculation

Every step has a specific purpose. Let's calculate a complete example.

Step 1 — Start With the Inputs

Suppose our neuron receives two input values.

x₁ = 2
x₂ = 3

Think of these simply as two pieces of information entering the neuron.

Step 2 — Give Each Input a Weight

Each input has a corresponding weight.

w₁ = 0.5
w₂ = 0.4

Now multiply each input by its weight.

x₁ × w₁
2 × 0.5 = 1.0

x₂ × w₂
3 × 0.4 = 1.2

We now have two weighted values:

Step 3 — Add the Weighted Values

Now add the results together.

1.0 + 1.2 = 2.2

Our weighted sum is therefore:

Step 4 — Add the Bias

Now suppose the neuron has a bias of:

b = 1

Add the bias to the weighted sum.

2.2 + 1 = 3.2

So the value before the activation function is:

The Formula

Instead of writing every calculation separately, we can represent the calculation with one formula:

z = (x₁ × w₁) + (x₂ × w₂) + b

Using our values:

z = (2 × 0.5) + (3 × 0.4) + 1

z = 1.0 + 1.2 + 1

z = 3.2

Step 5 — Apply an Activation Function

The value we calculated, 3.2, is normally passed through an activation function.

The activation function determines how the neuron's value is transformed.

We will study activation functions in detail in Lesson 3 — Activation Functions.

Complete Example From Start to Finish

Let's put everything together.

Inputs:

x₁ = 2
x₂ = 3


Weights:

w₁ = 0.5
w₂ = 0.4


Bias:

b = 1

First calculate the weighted inputs:

2 × 0.5 = 1.0
3 × 0.4 = 1.2

Add them:

1.0 + 1.2 = 2.2

Add the bias:

2.2 + 1 = 3.2

Finally, pass the result through the activation function.

Example With Three Inputs

Real neural network neurons can receive many inputs. Suppose we have three:

x₁ = 2
x₂ = 4
x₃ = 5

Their weights are:

w₁ = 0.3
w₂ = 0.5
w₃ = 0.2

And the bias is:

b = 1

Calculate each weighted input:

2 × 0.3 = 0.6
4 × 0.5 = 2.0
5 × 0.2 = 1.0

Add them:

0.6 + 2.0 + 1.0 = 3.6

Add the bias:

3.6 + 1 = 4.6

Calculate a Neuron With Python

We can perform the same calculation using simple Python.

x1 = 2
x2 = 3

w1 = 0.5
w2 = 0.4

bias = 1

z = (x1 * w1) + (x2 * w2) + bias

print(z)

Output:

3.2

The Python code is doing exactly the same calculation we performed manually.

Turn It Into a Function

We can make the calculation reusable by putting it inside a function.

def neuron_output(x1, x2, w1, w2, bias):
    return (x1 * w1) + (x2 * w2) + bias


result = neuron_output(2, 3, 0.5, 0.4, 1)

print(result)

Output:

3.2

This function represents the basic mathematical calculation performed by a very simple neuron.

What Is the Neuron Actually Learning?

The neuron is not normally learning the input values. The input data is provided to it.

During training, the model learns better values for its parameters, especially the weights and bias.

Why Do Different Weights Matter?

Consider two inputs with different weights.

Input 1 = 10
Weight 1 = 0.1

Input 2 = 10
Weight 2 = 0.9

Their contributions are very different:

10 × 0.1 = 1
10 × 0.9 = 9

Even though both inputs are 10, their contributions to the neuron are different because their weights are different.

What If a Weight Is Negative?

A weight can also be negative.

Input = 5
Weight = -0.5

5 × -0.5 = -2.5

A negative weight contributes negatively to the weighted sum.

This is why it is more accurate to say that a weight controls the direction and strength of an input's influence, rather than simply saying that it represents importance.

The Complete Neuron

Inputs
   ↓
Multiply each input by its weight
   ↓
Add weighted values
   ↓
Add bias
   ↓
Apply activation function
   ↓
Neuron output

From One Neuron to a Neural Network

One neuron is only a small part of a neural network. Many neurons are connected together in layers.

Each neuron performs its own calculation, and the outputs are passed through the network until the final prediction is produced.

The Big Picture

This is the basic operation that happens repeatedly inside a neural network.