Dropout
Domain: AI Fundamentals
🎯 Motivation
Randomly turns off some neurons during training so the network does not rely too much on any one path.
📋 Concrete Example
During each training step, 20% of neurons are randomly dropped. At inference, all neurons are used.
🔗 Analogy
Like a sports coach benching random players during practice so the team learns to work together, not depend on one star.