by-liu/MbLS
Code of our method MbLS (Margin-based Label Smoothing) for network calibration. To Appear at CVPR 2022. Paper : https://arxiv.org/abs/2111.15430
This project helps computer vision practitioners train more reliable image classification models. It takes your existing image datasets and model architectures as input, then applies a technique called Margin-based Label Smoothing (MbLS) during training. The output is a better-calibrated image classification model, meaning its predicted probabilities are more accurate reflections of its confidence, which is crucial for high-stakes applications. Scientists, engineers, or anyone building image-based decision systems would use this.
No commits in the last 6 months.
Use this if you need your image classification models to not only be accurate in their predictions but also trustworthy in their stated confidence levels, especially in fields like medical imaging or autonomous driving.
Not ideal if your primary concern is solely maximizing classification accuracy without regard for the reliability of the predicted probabilities.
Stars
50
Forks
7
Language
Python
License
MIT
Category
Last pushed
Nov 12, 2022
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/by-liu/MbLS"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
facebookincubator/MCGrad
MCGrad is a scalable and easy-to-use tool for multicalibration. It ensures your ML model...
dholzmueller/probmetrics
Post-hoc calibration methods and metrics for classification
gpleiss/temperature_scaling
A simple way to calibrate your neural network.
yfzhang114/Generalization-Causality
关于domain generalization，domain adaptation，causality，robutness，prompt，optimization，generative...
Affirm/splinator
Splinator: probabilistic calibration with regression splines