Forward Pass

Definition

A forward pass is the process of sending input through a network to produce output.

Basic pattern:

input -> layer 1 -> layer 2 -> output

Each layer transforms the values and passes them forward.

A forward pass computes the network's output.

---

Why It Is Called Forward

It is called forward because values move from the input side toward the output side.

input layer
-> hidden layer
-> output layer

The computation follows the direction of the network structure.

Forward means input-to-output computation.

---

What Happens During a Forward Pass

At each layer:

1. take incoming values
2. compute weighted sums
3. add biases
4. apply activation functions
5. pass outputs to the next layer

This repeats until the final output layer is reached.

A forward pass is repeated layer transformation.

---

Forward Pass Uses Current Parameters

A forward pass uses the parameters as they currently are.

Example:

current weights and biases -> output

If parameters change, the same input can produce a different output.

old parameters -> output A
new parameters -> output B

The forward pass shows what the current network computes.

---

Forward Pass Before Loss

Loss can be computed only after the network produces an output.

forward pass -> output -> loss computation

So the forward pass comes before loss measurement.

First compute the output; then measure how wrong it is.

---

Final Summary

• A forward pass sends input through the network.
• Values move from input layer to output layer.
• Each layer transforms the values.
• The forward pass uses the current parameters.
• Changing parameters can change the output.
• Loss is computed after the forward pass.