What is learning gaps?

Example: You map your understanding of machine learning and notice that 'gradient descent' connects to 'loss functions' and 'neural networks,' but 'regularization' sits as a near-orphan with a single weak link. You've read about regularization. You can define it. But your graph reveals the truth: you don't understand how it relates to overfitting, bias-variance tradeoff, or model selection. The gap between nodes isn't a formatting problem. It's a learning problem — and now you know exactly where to study next.

This concept is part of Knowledge Graphs in the How to Think curriculum, which builds the epistemic infrastructure for knowledge graphs.