Model Quality -- ML in Production Course @ CMU -- Lecture 5
Автор: Christian Kästner
Загружено: 2026-02-18
Просмотров: 41
Описание:
This is the fifth lecture of the Machine Learning in Production course (17-645/11-695) at Carnegie Mellon University by Claire Le Goues and Christian Kaestner, discussing pitfalls of traditional model evaluations, and more broadly how to design good measures. Other approaches to model testing beyond accuracy evaluations will be discussed in a subsequent lecture.
Course website: https://mlip-cmu.github.io/s2026/
00:00 Preliminaries
02:02 Case Study: Cancer Prognosis
03:44 Terminology
05:30 Measuring and Comparing Accuracy
12:03 The Costs of Mistakes
14:41 Detour: Measurement
27:51 Breakout: Defining Measures
41:44 Measurement Validity
43:51 Pitfalls in Measuring Model Quality
Повторяем попытку...
Доступные форматы для скачивания:
Скачать видео
-
Информация по загрузке:
![Почему взрываются батарейки и аккумуляторы? [Veritasium]](https://imager.clipsaver.ru/a3-3R9zwyGY/max.jpg)
