Short Answer

In one to three sentences, explain what upper bound a 5% optimization-algorithm error share places on possible improvements.

Question: If Optimization Verification shows only 5% of errors are due to the optimization algorithm, what is the realistic upper bound on error reduction achievable by improving that algorithm, and why?

Sample answer: The realistic upper bound is about 5% error reduction, because that is the maximum share of errors attributable to the optimization algorithm. Even a perfect optimization procedure cannot fix errors caused by the scoring function.

Key points:

  • Upper bound is about 5% error reduction
  • Bound comes from the error share attributed to the optimization algorithm
  • Perfecting the optimization algorithm cannot fix scoring function errors

Rubric: Full credit requires stating the ~5% ceiling and explaining that it reflects the error share attributed to the optimization algorithm.

0

1

Updated 2026-07-10

Contributors are:

Who are from:

Tags

Machine Learning

Deep Learning

Supervised Learning

Dive into Deep Learning @ D2L

Data Science

Machine Learning Strategy

Machine Learning Yearning @ DeepLearning.AI