What is over-parameterization in high dimensions?

Updated May 15, 2026

Short answer

When models have more parameters than data points.

Deep explanation

This increases risk of memorization rather than learning general patterns.

Real-world example

Deep networks with millions of parameters.

Common mistakes

  • Assuming complexity always improves accuracy.

Follow-up questions

  • Why is it still used?
  • What controls it?

More Curse of Dimensionality interview questions

View all →