Intro to Out-of-Distribution Detection
The below tutorial from Sharon Li, an Assistant Professor in the Department of Computer Sciences at the University of Wisconsin-Madison, introduces a pervasive problem faced by many machine learning systems deployed in the wild — out-of-distribution data.
Out-of-distribution data, often overlooked but immensely consequential, poses a significant threat to the reliability and efficacy of machine learning models. Through Sharon’s presentation, viewers gain a comprehensive understanding of this complex phenomenon and its potential ramifications on predictive accuracy.
Check out the video below to learn more about this problem and the cutting-edge methods you can equip yourself with to prevent inaccurate model predictions.
Prerequisites
- Comfort with supervised classification and the idea of a training distribution. The Intro to Machine Learning with Sklearn workshop is enough background.
- Some deep learning helps for the later methods; see Intro to Deep Learning with PyTorch.
Estimated time to complete
About an hour for the tutorial video. Use the Improve this page functionality to update this estimate!
Comments