Does one successful bounded minimization establish the global best hyperparameter for a multimodal scalar loss?
MathematicsLevel: IntermediateScalar optimization
Does one successful bounded minimization establish the global best hyperparameter for a multimodal scalar loss?
Hints
- Multiple valleys may exist.
- The initial interval influences search.
Answer
No. It is a local method. Explore the domain, use multiple brackets or a justified global strategy, then compare feasible objective values.
Source and licence: SciPy — scalar and multivariate minimization · BSD-3-Clause · answer: Quantoros editorial