Explain 'Global vs Local' Minima.

Explain 'Global vs Local' Minima.

Approach: Compare the absolute lowest point of a function with the lowest point in a specific neighborhood.

A local minimum is the lowest point in a specific region of a function. The global minimum is the absolute lowest point of the entire function. Non-convex functions (like deep neural networks) have many local minima.

Follow-up: Does SGD help escape local minima?

Key concepts: convexity, non-convex, optimization.