Glossary term
Glossary term
Safety and Alignment
An algorithm that balances two goals:
The need to build the most predictive model (for example, lowest loss).
The need to keep the model as simple as possible (for example, strong regularization).
For example, a function that minimizes loss+regularization on the training set is a structural risk minimization algorithm.
Contrast with empirical risk minimization.
For example, a function that minimizes loss+regularization on the training set is a structural risk minimization algorithm.
Contrast with empirical risk minimization.
Created for this library
A research team uses structural risk minimization principles to choose model complexity that balances training fit and generalization.
An ML team applies SRM ideas when selecting model complexity that balances training fit and held-out performance.
A risk modeling team frames its model complexity choice in SRM terms when explaining training and generalization trade-offs to reviewers.
Definition source: Google for Developers Machine Learning Glossary | Creative Commons Attribution 4.0 License