OOD Machine Learning: Detection, Robustness, and Generalization
A curated collection of papers, tutorials, books, benchmarks and open-source libraries on out-of-distribution detection, robustness and generalization in machine learning.
This resource organizes material on out-of-distribution, or OOD, problems in machine learning into three related areas: OOD detection, which aims to flag inputs that deviate from a model's training distribution for review or fallback handling; OOD robustness, which is about keeping performance reliable when test data differs from training data; and OOD generalization, which is about extrapolating to novel domains without retraining.
It is organized into sections covering researchers with sustained contributions to the field, articles, talks, benchmarks, libraries and datasets, surveys, theses, and a papers section further split into OOD detection, OOD robustness, OOD generalization and other related work.
It is released under the CC0-1.0 license.