Run One Forward Pass by Hand
Compute and explain a simple forward pass using inputs, weights, bias, activation, and output.
A simple renewal-risk neuron receives two binary inputs: low usage = 1 and unresolved ticket = 1. Weight for low usage is 0.7. Weight for unresolved ticket is 0.4. Bias is -0.5. Activation is ReLU, which outputs zero for negative values and keeps positive values. Forward pass = multiply inputs by weights, add bias, apply activation, send the result forward. The common trap is to describe the neuron as voting or thinking. It is doing a numerical transformation with learned parameters. Multiply low usage 1 x 0.7 = 0.7; unresolved ticket 1 x 0.4 = 0.4 Weights control the strength…
Sign up free — one personalized lesson every day, matched to your role and goals.
Already have an account? Sign in