How Forward Propagation Works
Raw Features Go In
The network sees numbers, not pictures — each feature is a measurement
The network doesn't see a photo — it sees numbers. Each feature is a measurement scaled to 0–1. High values mean "strongly present." These numbers are the only thing the network has to work with.
Step 1 of 6: Raw Features Go In
The network sees numbers, not pictures — each feature is a measurement
The network doesn't see a photo — it sees numbers. Each feature is a measurement scaled to 0–1. High values mean "strongly present." These numbers are the only thing the network has to work with.
Step 2 of 6: Weights Shape the Signal
Every connection multiplies the input by a learned weight
Every connection has a weight — a number that amplifies or dampens the signal. Tap any neuron to see the weights feeding into it.
Step 3 of 6: Neurons Sum It Up
Each neuron collects weighted signals and adds them together
Each neuron is a collector. It takes every weighted input and sums them into one number — the weighted sum (z). This raw sum can be any real number. The activation function comes next.
Step 4 of 6: The Activation Function
Sigmoid squashes any number into a smooth 0–1 range
The sigmoid function squashes any number into a clean 0–1 range. Big positive → nearly 1. Big negative → nearly 0. This gives the neuron a smooth on/off switch — not binary, but a confidence level.
Step 5 of 6: Layer by Layer
Watch data ripple forward through the entire network
Each layer transforms its inputs with weights → sum → sigmoid, then passes the result forward. The features get progressively more abstract — from "pointy ears" to "cat-like patterns" to the final answer. No layer can peek ahead or look back.
Step 6 of 6: The Prediction
The output layer speaks — the network has made its guess