Skip to content
BENIBIX Learn

Weight (Neural Network)

AI📈 Machine LearningNeural Networks

Also known as Connection Weight, Synaptic Weight

A number on each link in a network. It sets how much an input counts.

Think of it like

Think of it like a volume knob on each link. Turn it up and that input counts more.

Example

In Emma's house price network, one weight links size to the next layer. Training pushes it up to 2.4, because size matters a lot for price.

Why it matters

The weights hold what a network has learned. When you download a model, the weights are most of what you get.

Definition

A weight is a number on a link between two neurons. It says how much one neuron affects the other. A big weight makes an input count a lot, and a weight near zero makes it count very little. A negative weight pushes the other way. Training is mostly the work of finding good weights.
The numbers on the arrows are weights. A bigger number means that input counts for more.

Common mistake

Thinking you can read one weight and know why the model decided. In a big network, meaning is spread over millions of weights.

Key takeaways

  1. 1
    On every link. Between two neurons.
  2. 2
    Sets how much. Big counts more, near zero counts less.
  3. 3
    Learned in training. Training finds good values.

Learn it properly