Activation Function
Domain: AI Fundamentals
🎯 Motivation
Adds non-linearity so the network can learn complex patterns, not just straight-line rules.
📋 Concrete Example
ReLU turns negative inputs to zero and keeps positives. Sigmoid squashes values to between 0 and 1.
🔗 Analogy
Like a gate that only opens if you push hard enough - without it, the network would only draw straight lines.