Misconceptions in Distribution Shift
Daniel and Zachary discuss the common misconceptions surrounding distribution shift in machine learning, highlighting the confusion caused by unqualified claims and the lack of understanding about methods and assumptions. They emphasize the need for more rigorous evaluation and the potential failure of benchmark-driven domain adaptation methods.In this clip
From this podcast
The Gradient
Zachary Lipton: Where Machine Learning Falls Short
Related Questions