New approach improves adversarial robustness without sacrificing natural generalization.
problem Balancing adversarial robustness and natural generalization in machine learning.
method Friendly adversarial training (FAT) using early-stopped PGD to find least adversarial data.
result Early-stopped PGD achieves adversarial robustness without compromising natural generalization.
Proposes a new adversarial model to avoid accuracy vs. adversarial accuracy tradeoff.
problem Inherent tradeoff between accuracy and adversarial accuracy in existing adversarial robustness definitions.
method Introduces Voronoi-epsilon adversary that balances perturbation constraints.
result Voronoi-epsilon adversary avoids accuracy vs. adversarial accuracy tradeoff even with large ε. SPAT improves adversarial robustness by preserving semantics in adversarial training.
problem Adversarial examples often have different semantics than original data, introducing unintended biases.
method Semantics-preserving adversarial training (SPAT) that encourages pixel perturbation shared among all classes.
result SPAT improves adversarial robustness and achieves state-of-the-art results in CIFAR-10 and CIFAR-100.
More training data can hurt the generalization of adversarially robust models.
problem The challenge of balancing adversarial robustness and generalization in machine learning models.
method Investigation of three regimes based on adversary strength and empirical studies on various models.
result More training data can hurt the generalization of adversarially robust models in different regimes.
Paper proposes a new black-box adversarial attack using normalizing flows.
problem Adversarial vulnerability of deep neural networks.
method Proposes a novel black-box adversarial attack using normalizing flows.
result Demonstrates competitive performance against well-known black-box adversarial attack methods.
We present a data-driven framework called generative adversarial privacy (GAP). Inspired by recent advancements in generative adversarial networks (GANs), GAP allows the data holder to learn the privatization mechanism directly from the data. Under GAP, finding the optimal privacy mechanism is formulated as a constrain…
Paper tackles robustness in adversarial noise with a meta-optimizer.
problem Sensitivity to adversarial noise hinders machine learning deployment.
method Meta-optimizer learns to robustly optimize models using adversarial examples.
result Meta-optimizer transfers adversarial knowledge to new models without generating new examples.
Adversarial training achieves optimal test error for shallow networks.
problem Achieving optimal adversarial test error for general data distributions.
method Applying new Rademacher complexity bounds and properties of optimal adversarial predictors.
result Adversarial training can achieve optimal adversarial test error for general data distributions.
Paper proposes a method to train robust neural networks without labeled data.
problem Training robust neural networks without class labels.
method Adversarial contrastive learning framework using unlabeled data.
result Robust Contrastive Learning (RoCL) achieves comparable robust accuracy to supervised methods and significantly improved robustness.
This paper benchmarks time-series adversarial defenses and attacks.
problem Adversarial attacks on time-series data.
method Detailed benchmarking of adversarial defense methods in the L∞ threat model. result Adversarial defenses offer robustness against both strong white-box and black-box attacks.
New method improves adversarial learning with unlabeled data.
problem Poor quality of pseudo labels on unlabeled data.
method Robust Co-training (RCT) using deep co-training.
result RCT significantly outperforms baselines in adversarial robustness.
Adversarial training can lead to overfitting without compromising robustness.
problem Explaining benign overfitting in adversarially robust linear classification.
method Theoretical analysis and numerical experiments on adversarial training.
result Adversarially trained linear classifiers can achieve near-optimal risks despite overfitting noisy data.
Paper analyzes adversarial training's performance in binary classification.
problem Understanding the generalization performance of adversarial training.
method Derives precise theoretical predictions for adversarial training performance.
result Provides exact asymptotics for test errors of adversarial training.
Paper discusses imperceptible attacks on tabular data models.
problem Security of machine learning models on tabular data.
method Formalizes imperceptibility of adversarial attacks, proposes an approach to generate imperceptible adversarial examples.
result Demonstrates ability to generate imperceptible adversarial examples with high fooling rate.
This paper explores tradeoffs between standard and adversarial risks in distributionally adversarial training.
problem Understanding the impact of adversarial training on standard risk and adversarial risk.
method Study of distributionally adversarial training with different learning settings and models.
result Derives Pareto-optimal tradeoff curves between standard and adversarial risks.
Linear classifiers can resist adversarial attacks on Gaussian data.
problem Adversarial attacks on high-dimensional data.
method Adversarial training of linear classifiers on Gaussian data.
result Linear classifiers can resist adversarial attacks on Gaussian data.
ACL improves robustness with unlabeled data, and we analyze its generalization using Rademacher complexity.
problem Improving robustness of deep networks against adversarial attacks using unlabeled data.
method We analyze the generalization performance of Adversarial Contrastive Learning (ACL) using Rademacher complexity.
result The average adversarial risk of the downstream tasks can be upper bounded by the adversarial unsupervised risk of the upstream task.
DIP-FAT improves adversarial training by diversifying perturbations.
problem Adversarial examples fool deep neural networks, leading to overfitting and poor performance.
method DIP-FAT uses random directions to diversify perturbations in adversarial training.
result DIP-FAT reduces overfitting and improves clean data accuracy.
New uniqueness concept for adversarial Bayes classifier.
problem Understanding adversarial Bayes classifiers in binary classification.
method Developed a new notion of uniqueness and analyzed it for a family of one-dimensional data distributions.
result Improved regularity of adversarial Bayes classifiers as perturbation radius increases.
The paper evaluates machine learning cyber defenses using log data against adversarial attacks.
problem Evaluating the robustness of machine learning cyber defenses against adversarial attacks.
method Developed a testing framework using deep reinforcement learning and adversarial natural language processing.
result Higher dropout levels increase robustness, with 90% dropout probability showing the highest robustness.
Most previous works usually explained adversarial examples from several specific perspectives, lacking relatively integral comprehension about this problem. In this paper, we present a systematic study on adversarial examples from three aspects: the amount of training data, task-dependent and model-specific factors. Pa…
This research examines how data transformations affect adversarial robustness in recurrent neural networks.
problem Adversarial examples reduce machine learning accuracy, especially in high-dimensional datasets.
method Analysis of feature selection, dimensionality reduction, and trend extraction techniques on recurrent neural networks.
result Data transformations may increase vulnerability to adversarial samples, but only if they approximate intrinsic dimensionality and maintain manifold coverage.
Paper examines adversarial attacks on event cause analysis in power grids.
problem Adversarial attacks manipulate data to mislead event classifiers.
method Investigated adversarial attacks on CNN-based event cause analysis.
result Demonstrated adversaries can misclassify events through stealthy data manipulations.
DAD++ improves test-time adversarial defense without access to training data.
problem Improving adversarial robustness of pre-trained models without access to training data.
method Test-time Data-free Adversarial Defense (DAD++) with detection and correction schemes.
result DAD++ significantly improves adversarial robustness with minimal clean accuracy drop.
Nowadays more and more data are gathered for detecting and preventing cyber attacks. In cyber security applications, data analytics techniques have to deal with active adversaries that try to deceive the data analytics models and avoid being detected. The existence of such adversarial behavior motivates the development…
Paper shows adversarial training can be fooled by new type of noise.
problem Adversarial training can be fooled by new types of noise.
method Designing ADVIN, a new type of inducing noise.
result ADVIN can degrade adversarial training robustness by 99.9%.
Enhances adversarial robustness with unlabeled out-of-domain data.
problem Improving robustness of models against adversarial attacks.
method Leveraging unlabeled data from multiple domains to bridge the sample complexity gap in adversarial robustness.
result Better adversarial robustness achieved when unlabeled data comes from a shifted domain.
MMD test detects adversarial attacks by addressing kernel limitations and non-independence issues.
problem MMD test's failure to detect adversarial attacks.
method Replaced Gaussian kernel with deep kernel, maximized test power, and used wild bootstrap for non-independence.
result MMD test is aware of adversarial attacks.
Study analyzes adversarial training dynamics without data distribution assumptions.
problem Understanding training dynamics of adversarial training without data distribution assumptions.
method Mean field theory approach to analyze adversarial training in random deep neural networks.
result Upper bounds of adversarial loss derived empirically and theoretically.
The paper identifies potential adversarial samples near decision boundaries of neural networks.
problem Vulnerability of deep neural networks to small perturbations of inputs.
method Developed a method to explore near decision boundaries of trained classifiers to identify potential adversarial samples.
result Potential adversarial samples represent only 61% of the test data but cover more than 82% of adversarial samples produced by iFGSM and 92% of those by DeepFool on CIFAR10.
Recent studies have highlighted adversarial examples as a ubiquitous threat to different neural network models and many downstream applications. Nonetheless, as unique data properties have inspired distinct and powerful learning principles, this paper aims to explore their potentials towards mitigating adversarial inpu…
Neural networks are vulnerable to small adversarial perturbations. Existing literature largely focused on understanding and mitigating the vulnerability of learned models. In this paper, we demonstrate an intriguing phenomenon about the most popular robust training method in the literature, adversarial training: Advers…
Adversarial robustness has become a central goal in deep learning, both in the theory and the practice. However, successful methods to improve the adversarial robustness (such as adversarial training) greatly hurt generalization performance on the unperturbed data. This could have a major impact on how the adversarial …
This paper improves adversarial robustness of deep learning models.
problem Vulnerability of machine learning models to adversarial perturbations.
method Analyzes adversarial training for linear regression and neural networks, incorporating L1 penalty.
result Incorporating L1 penalty leads to consistent adversarially robust estimation in high-dimensional settings.
Paper generates natural adversarial examples for hyperspectral data.
problem Creating adversarial examples for black-box models.
method Modified Wasserstein GAN reweights true data distribution.
result Successfully generates adversarial hyperspectral signatures.
Recent work on adversarial attack has shown that Projected Gradient Descent (PGD) Adversary is a universal first-order adversary, and the classifier adversarially trained by PGD is robust against a wide range of first-order attacks. It is worth noting that the original objective of an attack/defense model relies on a d…
Deep learning models are vulnerable to various adversarial manipulations of their training data, parameters, and input sample. In particular, an adversary can modify the training data and model parameters to embed backdoors into the model, so the model behaves according to the adversary's objective if the input contain…
Neural network robustness has recently been highlighted by the existence of adversarial examples. Many previous works show that the learned networks do not perform well on perturbed test data, and significantly more labeled data is required to achieve adversarially robust generalization. In this paper, we theoretically…
This paper explores the use of adversarial examples in training speech recognition systems to increase robustness of deep neural network acoustic models. During training, the fast gradient sign method is used to generate adversarial examples augmenting the original training data. Different from conventional data augmen…
Game theory helps machine learn better from adversarial queries.
problem Adversarial evasion in machine learning prediction.
method Repeated Bayesian Sequential Game to balance classifier selection and query type.
result Learner selects appropriate classifier for clean vs. adversarial queries.
New approach detects adversarial samples with certifiable guarantees.
problem Adversarial samples can trick CNNs, posing a threat.
method Certifiable Taboo Trap (CTT) approach to detect adversarial inputs.
result CTT outperforms existing defenses on various lp norms. Framework to mitigate adversarial attacks by allowing classifiers to abstain.
problem Vulnerability of classifiers to adversarial examples.
method Adversarial training with a rejection option.
result ATRO framework improves classifier reliability against adversarial attacks.
Adversarial examples are a pervasive phenomenon of machine learning models where seemingly imperceptible perturbations to the input lead to misclassifications for otherwise statistically accurate models. We propose a geometric framework, drawing on tools from the manifold reconstruction literature, to analyze the high-…
Adversarial noises are linearly separable for random neural networks.
problem The challenge of adversarial examples in neural networks.
method Theoretical proof and empirical evidence for two-layer networks with random initialization and neural tangent kernel setup.
result Adversarial noises are linearly separable with corresponding labels.
The paper analyzes adversarial training effects on classification accuracy.
problem Understanding adversarial training's impact on standard and robust accuracy.
method Derived precise statistical analysis for binary classification problems with Gaussian data.
result Theoretical explanation of standard and robust accuracy trends for adversarial training.
This paper introduces a new method to deceive causal structure learning by omitting data.
problem Deceiving causal structure learning algorithms with incompletely observed data.
method Adversarial missingness attack to bias the learned causal structures.
result Theoretical and practical attack mechanisms are developed for various SCMs.
Data augmentation impacts adversarial risk; careful application recommended.
problem Understanding how data augmentation affects adversarial risk in deep learning.
method Empirical analysis using three measures of adversarial risk.
result Data augmentation does not always improve adversarial risk; augmented data influences models more.
I describe an optimal control view of adversarial machine learning, where the dynamical system is the machine learner, the input are adversarial actions, and the control costs are defined by the adversary's goals to do harm and be hard to detect. This view encompasses many types of adversarial machine learning, includi…