Deep Neural Networks (DNNs) have revolutionized a multitude of machine learning applications but are notorious for their susceptibility to adversarial attacks. Our project introduces CARE (Certifiably Robust leArning with REasoning), aiming to enhance the robustness of DNNs by integrating them with reasoning abilities. This pipeline consists of two primary components:
-
Learning Component: Utilizes CLIP model for different semantic predictions, e.g., recognizing if an input image contains something furry.
-
Reasoning Component: Employs probabilistic graphical models like Markov Logic Networks (MLN) to apply domain-specific knowledge and logic reasoning to the learning process.
conda create -n name care python=3.9
conda activate care && pip install -r requirements.txt
from defense import load_care_model
# the input is a tensor with shape N * 3 * 32 * 32 in range [0,1]
# for adv attack （although our model is trained with Gaussian Augmentation instead of adversarial training）
model = load_care_model().cuda()
# for ceritification with randomized smoothing, you need to specify the noise_sd
model = load_care_model(noise_sd = 0.25).cuda()
model = model.eval()
# And you can use model.predict(x: torch.tensor, n: int, alpha: float, batch_size: int)
# or model.certify(x: torch.tensor, n0: int, n: int, alpha: float, batch_size: int)
# to do the predicition or certification
If you find our work useful, please consider citing our paper:
@inproceedings{zhang2023care,
title={CARE: Certifiably Robust Learning with Reasoning via Variational Inference},
author={Zhang, Jiawei and Li, Linyi and Zhang, Ce and Li, Bo},
booktitle={2023 IEEE Conference on Secure and Trustworthy Machine Learning (SaTML)},
pages={554--574},
year={2023},
organization={IEEE}
}
If you have any questions or encounter any errors while running the code, feel free to contact jiaweiz@illinois.edu!