Glossary term
Glossary term
Training and Fine-Tuning
What you believe about the data before you begin training on it. For example, L2 regularization relies on a prior belief that weights should be small and normally distributed around zero.
Created for this library
A Bayesian risk model encodes prior beliefs about credit defaults from historical macroeconomic data alongside current applicant features.
A medical research team encodes prior beliefs about disease prevalence in a Bayesian model so estimates are sensible on small datasets.
A research team uses a Bayesian model with informative prior beliefs to combine domain knowledge with limited training data.
Definition source: Google for Developers Machine Learning Glossary | Creative Commons Attribution 4.0 License