Home

Awesome

Learning to Predict Trustworthiness with Steep Slope Loss

This project is based on ConfidNet . The code is tested with Python 3.8.5 and PyTorch 1.6.0/1.7.1 under Ubuntu 1804 equipped with Nvidia GTX 1080 ti graphics cards. More details can be found in our paper.

If you find this code useful for your research, please cite the following papers:

@incollection{NIPS2019_8556,
   title = {Addressing Failure Prediction by Learning Model Confidence},
   author = {Corbi\`{e}re, Charles and THOME, Nicolas and Bar-Hen, Avner and Cord, Matthieu and P\'{e}rez, Patrick},
   booktitle = {Advances in Neural Information Processing Systems 32},
   editor = {H. Wallach and H. Larochelle and A. Beygelzimer and F. d\textquotesingle Alch\'{e}-Buc and E. Fox and R. Garnett},
   pages = {2902--2913},
   year = {2019},
   publisher = {Curran Associates, Inc.},
   url = {http://papers.nips.cc/paper/8556-addressing-failure-prediction-by-learning-model-confidence.pdf}
}

@inproceedings{Luo_NeurIPS_2021,
  title={Learning to Predict Trustworthiness with Steep Slope Loss},
  author={Luo, Yan and Wong, Yongkang and Kankanhalli, Mohan and Zhao, Qi},
  booktitle={Advances in Neural Information Processing Systems},
  year={2021}

Installation

  1. Clone the repo.

  2. Install this repository and the dependencies using pip.

  3. Configure the root folder in the file predict_trustworthiness_smallscale/confidnet/train.py, e.g.,

import sys
sys.path.append('/home/yluo/project/python/oracle_learning/predict_trustworthiness_smallscale')

Datasets

MNIST and CIFAR-10 datasets are managed by Pytorch dataloader. First time you run a script, the dataloader will download the dataset in predict_trustworthiness_smallscale/confidnet/data/DATASETNAME-data.

Download the official pre-trained ConfidNet. Extract the pre-trained models to

predict_trustworthiness_smallscale/confidnet/pretrained/mnist_pretrained/confidnet/
predict_trustworthiness_smallscale/confidnet/pretrained/cifar10_pretrained/confidnet/

Running the code

Execute the following command for training:

$ cd predict_trustworthiness_smallscale/confidnet
$ ./scripts/train.sh

The resulting output files can be found in

experiments/mnist_steep/logs.csv
experiments/cifar10_steep/logs.csv

Pre-trained models

Fine-tuned models with the steep slope loss on MNIST and CIFAR-10 are available via this link.