fregu856/regression_uncertainty

Official implementation of "How Reliable is Your Regression Model's Uncertainty Under Real-World Distribution Shifts?", TMLR 2023.

27
/ 100
Experimental

This project helps machine learning researchers and practitioners understand how reliable their regression models are when encountering new, unexpected data. It takes common image-based regression datasets and evaluates various uncertainty estimation methods to see if they accurately reflect when the model is unsure about its predictions, especially under different real-world scenarios. The output is a benchmark showing which methods remain trustworthy even when the data changes.

No commits in the last 6 months.

Use this if you are developing or using regression models that need to provide reliable uncertainty estimates, particularly when deployed in environments where data characteristics might shift over time, such as in medical imaging or industrial inspection.

Not ideal if you are looking for a plug-and-play solution for a specific application without delving into the underlying research and model evaluation.

machine-learning-research model-reliability uncertainty-quantification image-regression distribution-shift
Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 6 / 25
Maturity 16 / 25
Community 5 / 25

How are scores calculated?

Stars

18

Forks

1

Language

Python

License

MIT

Last pushed

Mar 21, 2024

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/fregu856/regression_uncertainty"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.