Paper | Demonstration | Citation
This repository contains code for the TMLR Paper "Neural Clamping: Joint Input Perturbation and Temperature Scaling for Neural Network Calibration." The research demo of Neural Clamping can be found at NCTV: Neural Clamping Toolkit and Visualization for Neural Network Calibration, which earlier appeared in AAAI 2023 Demo Track.
Authors: Yung-Chen Tang, Pin-Yu Chen, and Tsung-Yi Ho.
What the differnet between Confidence and Accuracy?
Quick Start by running the following code!
# !pip install -q git+https://github.com/yungchentang/NCToolkit.git
from neural_clamping.nc_wrapper import NCWrapper
from neural_clamping.utils import load_model, load_dataset, model_classes, plot_reliability_diagram
# Load model
model = load_model(name='ARCHITECTURE', data='DATASET', checkpoint_path='CHECKPOINT_PATH')
num_classes = model_classes(data='DATASET')
# Dataset loader
valloader = load_dataset(data='DATASET', split='val', batch_size="BATCH_SIZE")
testloader = load_dataset(data='DATASET', split='test', batch_size="BATCH_SIZE")
# Build Neural Clamping framework
nc = NCWrapper(model=model, num_classes=num_classes, ...)
# Calibrated using Neural Clamping
nc.train_NC(val_loader=valloader, epoch='EPOCH', ...)
# General Evaluation
nc.test_with_NC(test_loader=testloader)
# Visualization
bin_acc, conf_axis, ece_score = nc.reliability_diagram(test_loader=testloader, rd_criterion="ECE", n_bins=30)
plot_reliability_diagram(conf_axis, bin_acc)
If you find this helpful for your research, please cite our papers as follows:
@article{tang2024neural,
title={{Neural Clamping: Joint Input Perturbation and Temperature Scaling for Neural Network Calibration}},
author={Yung-Chen Tang and Pin-Yu Chen and Tsung-Yi Ho},
journal={Transactions on Machine Learning Research},
issn={2835-8856},
year={2024},
url={https://openreview.net/forum?id=qSFToMqLcq},
}
@inproceedings{hsiung2023nctv,
title={{NCTV: Neural Clamping Toolkit and Visualization for Neural Network Calibration}},
author={Lei Hsiung and Yung-Chen Tang and Pin-Yu Chen and Tsung-Yi Ho},
booktitle={Proceedings of the Thirty-Seventh AAAI Conference on Artificial Intelligence},
publisher={Association for the Advancement of Artificial Intelligence},
year={2023},
month={February}
}