
Koray KavukcuogluResearcher
Chief AI Architect
via Google
Researcher
Google Fellow at Google
Platt created methods that made an important early class of machine-learning models faster to train and better at expressing uncertainty. One of them, Platt scaling, converts a model's raw score into a useful probability, such as an 80% chance that a prediction is correct.
See something inaccurate or outdated?