Awesome

Conformal Prediction for Deep Classifier via Label Ranking

This repository is the official implementation of Conformal Prediction for Deep Classifier via Label Ranking at ICML'2024

How to Install

This code is built on the awesome toolbox TorchCP that you need to install first. We use Python 3.9, and TorchCP can be installed by:

pip install torchcp

Simply follow the instructions described here to install TorchCP as well as PyTorch. After that, other packages can be installed by:

pip install -r requirements.txt

Then, you are ready to go.

How to Run

Producing the prediction sets:

python main.py --dataset imagenet  --trials 10

with the following arguments:

• dataset: the name of the dataset.
• trials: the number of trials.

Citation

If you find this useful in your research, please consider citing:

@inproceedings{huang2023conformal,
title={Conformal Prediction for Deep Classifier via Label Ranking},
author={Jianguo Huang and Huajun Xi and Linjun Zhang and Huaxiu Yao and Yue Qiu and Hongxin Wei},
booktitle = {International Conference on Machine Learning (ICML)},
year = {2024}
}