next

Neural Networks (Basics)

Adjust the network structure and activation function. Press Forward Pass to recompute activations. Color intensity shows neuron activation magnitude. Edge thickness shows |weight|. Hover / click for details.

Ready

Concept

  • Neuron: weighted sum + bias
  • Edge (weight): strength & direction of influence
  • Activation: introduces non‑linearity
  • Forward pass: values propagate layer by layer

Hidden layer values use the chosen activation; output layer here is linear.

Layer Outputs


            

            

                
Tips

                
Change layer sizes or activation then press Forward Pass. Hover nodes for formulas; click a node to freeze highlight.