paper-with-me

홈 › Papers

Automatic Perturbation Analysis for Scalable Certified Robustness and Beyond

2020-02-28 · NeurIPS 2020 12 · Kaidi Xu, Zhouxing Shi, huan zhang, Yihan Wang, Kai-Wei Chang, Minlie Huang, Bhavya Kailkhura, Xue Lin, Cho-Jui Hsieh

Linear relaxation based perturbation analysis (LiRPA) for neural networks, which computes provable linear bounds of output neurons given a certain amount of input perturbation, has become a core component in robustness verification and certified defense. The majority of LiRPA-based methods focus on simple feed-forward networks and need particular manual derivations and implementations when extended to other architectures. In this paper, we develop an automatic framework to enable perturbation analysis on any neural network structures, by generalizing existing LiRPA algorithms such as CROWN to operate on general computational graphs. The flexibility, differentiability and ease of use of our framework allow us to obtain state-of-the-art results on LiRPA based certified defense on fairly complicated networks like DenseNet, ResNeXt and Transformer that are not supported by prior works. Our framework also enables loss fusion, a technique that significantly reduces the computational complexity of LiRPA for certified defense. For the first time, we demonstrate LiRPA based certified defense on Tiny ImageNet and Downscaled ImageNet where previous approaches cannot scale to due to the relatively large number of classes. Our work also yields an open-source library for the community to apply LiRPA to areas beyond certified defense without much LiRPA expertise, e.g., we create a neural network with a probably flat optimization landscape by applying LiRPA to network parameters. Our opensource library is available at https://github.com/KaidiXu/auto_LiRPA.

📄 PDF Abstract BibTeX arXiv:2002.12920

Code (7)

KaidiXu/auto_LiRPA 공식 구현 pytorch
Verified-Intelligence/alpha-beta-CROWN_vnncomp2024 pytorch
avrahamraviv/formal-verification-of-object-detection pytorch
huanzhang12/CROWN-IBP pytorch
huanzhang12/alpha-beta-CROWN pytorch
shizhouxing/Robustness-Verification-for-Transformers pytorch
verified-intelligence/alpha-beta-crown pytorch

Tasks

Quantization

Methods 이 논문이 사용한 방법론

Linear Layer A Linear Layer is a projection $\mathbf{XW + b}$.
Absolute Position Encodings Absolute Position Encodings are a type of position embeddings for [Transformer-based models] where positional encodings are…
Position-Wise Feed-Forward Layer 설명 없음
Concatenated Skip Connection A Concatenated Skip Connection is a type of skip connection that seeks to reuse features by concatenating them to new layers, allowing more information to be retained from…
Max Pooling Max Pooling is a pooling operation that calculates the maximum value for patches of a feature map, and uses it to create a downsampled (pooled) feature map. It is usually…
Global Average Pooling Global Average Pooling is a pooling operation designed to replace fully connected layers in classical CNNs. The idea is to generate one feature map for each corresponding…
Average Pooling 설명 없음
Convolution A convolution is a type of matrix operation, consisting of a kernel, a small matrix of weights, that slides over input data performing element-wise multiplication with the…

Similar Papers 제목 키워드 기반

UniCR: Universally Approximated Certified Robustness via Randomized Smoothing

2022-07-05 · Hanbin Hong, Binghui Wang, Yuan Hong

We study certified robustness of machine learning classifiers against adversarial perturbations. In particular, we propose the first universally approximated certified robustness (UniCR) framework, which can approximate …

Certified Robustness for Top-k Predictions against Adversarial Perturbations via Randomized Smoothing

2019-12-20 · ICLR 2020 1 · Jinyuan Jia, Xiaoyu Cao, Binghui Wang, Neil Zhenqiang Gong

It is well-known that classifiers are vulnerable to adversarial perturbations. To defend against adversarial perturbations, various certified robustness results have been derived. However, existing certified robustnesses…

Two Birds, One Stone: Achieving both Differential Privacy and Certified Robustness for Pre-trained Classifiers via Input Perturbation

2021-09-29 · Pengfei Tang, Wenjie Wang, Xiaolan Gu, Jian Lou 외

Recent studies have shown that pre-trained classifiers are increasingly powerful to improve the performance on different tasks, e.g, neural language processing, image classification. However, adversarial examples from at…

image-classificationImage Classification

Towards Universal Certified Robustness with Multi-Norm Training

2024-10-03 · Enyi Jiang, David S. Cheung, Gagandeep Singh

Existing certified training methods can only train models to be robust against a certain perturbation type (e.g. $l_\infty$ or $l_2$). However, an $l_\infty$ certifiably robust model may not be certifiably robust against…

Certified Adversarial Robustness with Additive Noise

2018-09-10 · NeurIPS 2019 12 · Bai Li, Changyou Chen, Wenlin Wang, Lawrence Carin

The existence of adversarial data examples has drawn significant attention in the deep-learning community; such data are seemingly minimally perturbed relative to the original data, but lead to very different outputs fro…

Adversarial AttackAdversarial Robustness