This repository serves as a comprehensive, curated resource for researchers and practitioners focused on Out-of-Distribution (OOD) detection, robustness, and generalization in machine learning. It aims to be a one-stop shop for papers, tutorials, benchmarks, libraries, and more, addressing the critical challenge of AI model failures when encountering data outside their training distribution.

How It Works

The repository is structured to cover the multifaceted field of OOD research, categorizing resources into OOD Detection, OOD Robustness, and OOD Generalization. It includes a vast collection of academic papers, surveys, talks, and articles, alongside practical resources like benchmarks, datasets, and open-source libraries. This curated approach allows users to quickly find relevant materials and stay updated on the latest advancements and methodologies in building reliable AI systems for open-world environments.

Quick Start & Requirements

This repository is a curated list of resources, not a runnable software package. No installation or specific requirements are needed to access the information.

Highlighted Details

• Extensive collection of papers from major ML conferences (NeurIPS, ICML, CVPR, ICLR, etc.) dating back to 2017.
• Includes dedicated sections for OOD detection, robustness, and generalization, with numerous sub-categories.
• Features curated lists of researchers, talks, surveys, and theses in the field.
• Provides links to various benchmarks, libraries, and datasets for practical experimentation.

Maintenance & Community

The repository is maintained by huytransformer. It appears to be a community-driven effort, with a significant number of papers and resources listed, suggesting active community contribution and interest in the field.

Licensing & Compatibility

The repository itself is not a software package and does not have a specific license. The linked resources (papers, code) are subject to their respective licenses.

Limitations & Caveats

As a curated list, the repository's content is dependent on the maintainer's and community's efforts to keep it up-to-date. The sheer volume of papers listed might require significant effort to navigate and digest.