Single Neuron Forward Pass
What you'll learn: How a single artificial neuron transforms its inputs into an output through weighted summation, bias addition, and activation.
The Core Mechanism
A single neuron performs a surprisingly simple two-step calculation. First, it computes a weighted sum of all its inputs, adds a bias term, then passes that result through an activation function to produce the final output.
Think of it like a voting committee: each input gets a vote (the input value) with different importance weights. The bias is like a default starting position before any votes are counted. The activation function then interprets whether the total score should trigger the neuron to "fire" or what strength of signal to send forward.
The Mathematical Recipe
For a single neuron with inputs x₁, x₂, ..., xₙ:
- Weighted sum + bias:
z = w₁x₁ + w₂x₂ + ... + wₙxₙ + b - Activation:
a = f(z)wherefis your activation function
The intermediate value z (sometimes called the pre-activation or logit) captures the linear combination. The activation function f introduces nonlinearity, producing the neuron's final output a.
Why This Matters
This two-step process is the fundamental building block of all neural networks. Every neuron in every layer performs this exact computation during forward propagation. The weights and bias are the parameters the network learns during training, while the activation function is a design choice you make when building the architecture.
Key Takeaway: A single neuron's forward pass multiplies inputs by learned weights, adds a bias, and applies an activation function—this simple operation, repeated across millions of neurons, enables deep learning's power.