Glossary term
Glossary term
Foundations
A function in which the region above the graph of the function is a convex set. The prototypical convex function is shaped something like the letter U. For example, the following are all convex functions:
In contrast, the following function is not convex. Notice how the region above the graph is not a convex set:
A strictly convex function has exactly one local minimum point, which is also the global minimum point. The classic U-shaped functions are strictly convex functions. However, some convex functions (for example, straight lines) are not U-shaped.
Click the icon for a deeper look at the math.
See Convergence and convex functions in Machine Learning Crash Course for more information.
Created for this library
A risk modeling team prefers loss functions that are convex functions because optimization is guaranteed to reach a single global minimum.
A pricing team uses logistic loss, a convex function, so its production training pipeline is robust to changes in initial parameters.
An ML platform team selects convex loss functions for production scorecards because they simplify model validation for risk audits.
Definition source: Google for Developers Machine Learning Glossary | Creative Commons Attribution 4.0 License