اشترك بالحزمة الذهبية واحصل على وصول غير محدود شمرا أكاديميا
تسجيل مستخدم جديدCombating Adversaries with Anti-Adversaries
80
0
0.0
(
0
)
نشر من قبل
Motasem Alfarra Alfarra M
تاريخ النشر
2021
مجال البحث
الهندسة المعلوماتية
والبحث باللغة
English
اسأل ChatGPT حول البحث
ﻻ يوجد ملخص باللغة العربية
Deep neural networks are vulnerable to small input perturbations known as adversarial attacks. Inspired by the fact that these adversaries are constructed by iteratively minimizing the confidence of a network for the true class label, we propose the anti-adversary layer, aimed at countering this effect. In particular, our layer generates an input perturbation in the opposite direction of the adversarial one, and feeds the classifier a perturbed version of the input. Our approach is training-free and theoretically supported. We verify the effectiveness of our approach by combining our layer with both nominally and robustly trained models, and conduct large scale experiments from black-box to adaptive attacks on CIFAR10, CIFAR100 and ImageNet. Our anti-adversary layer significantly enhances model robustness while coming at no cost on clean accuracy.
قيم البحث
اقرأ أيضاً
We prove novel algorithmic guarantees for several online problems in the smoothed analysis model. In this model, at each time an adversary chooses an input distribution with density function bounded above by $tfrac{1}{sigma}$ times that of the unifor
m distribution; nature then samples an input from this distribution. Crucially, our results hold for {em adaptive} adversaries that can choose an input distribution based on the decisions of the algorithm and the realizations of the inputs in the previous time steps. This paper presents a general technique for proving smoothed algorithmic guarantees against adaptive adversaries, in effect reducing the setting of adaptive adversaries to the simpler case of oblivious adversaries. We apply this technique to prove strong smoothed guarantees for three problems: -Online learning: We consider the online prediction problem, where instances are generated from an adaptive sequence of $sigma$-smooth distributions and the hypothesis class has VC dimension $d$. We bound the regret by $tilde{O}big(sqrt{T dln(1/sigma)} + dsqrt{ln(T/sigma)}big)$. This answers open questions of [RST11,Hag18]. -Online discrepancy minimization: We consider the online Komlos problem, where the input is generated from an adaptive sequence of $sigma$-smooth and isotropic distributions on the $ell_2$ unit ball. We bound the $ell_infty$ norm of the discrepancy vector by $tilde{O}big(ln^2!big( frac{nT}{sigma}big) big)$. -Dispersion in online optimization: We consider online optimization of piecewise Lipschitz functions where functions with $ell$ discontinuities are chosen by a smoothed adaptive adversary and show that the resulting sequence is $big( {sigma}/{sqrt{Tell}}, tilde Obig(sqrt{Tell} big)big)$-dispersed. This matches the parameters of [BDV18] for oblivious adversaries, up to log factors.
Poisoning attacks have emerged as a significant security threat to machine learning (ML) algorithms. It has been demonstrated that adversaries who make small changes to the training set, such as adding specially crafted data points, can hurt the perf
ormance of the output model. Most of these attacks require the full knowledge of training data or the underlying data distribution. In this paper we study the power of oblivious adversaries who do not have any information about the training set. We show a separation between oblivious and full-information poisoning adversaries. Specifically, we construct a sparse linear regression problem for which LASSO estimator is robust against oblivious adversaries whose goal is to add a non-relevant features to the model with certain poisoning budget. On the other hand, non-oblivious adversaries, with the same budget, can craft poisoning examples based on the rest of the training data and successfully add non-relevant features to the model.
Many machine learning image classifiers are vulnerable to adversarial attacks, inputs with perturbations designed to intentionally trigger misclassification. Current adversarial methods directly alter pixel colors and evaluate against pixel norm-ball
s: pixel perturbations smaller than a specified magnitude, according to a measurement norm. This evaluation, however, has limited practical utility since perturbations in the pixel space do not correspond to underlying real-world phenomena of image formation that lead to them and has no security motivation attached. Pixels in natural images are measurements of light that has interacted with the geometry of a physical scene. As such, we propose the direct perturbation of physical parameters that underly image formation: lighting and geometry. As such, we propose a novel evaluation measure, parametric norm-balls, by directly perturbing physical parameters that underly image formation. One enabling contribution we present is a physically-based differentiable renderer that allows us to propagate pixel gradients to the parametric space of lighting and geometry. Our approach enables physically-based adversarial attacks, and our differentiable renderer leverages models from the interactive rendering literature to balance the performance and accuracy trade-offs necessary for a memory-efficient and scalable adversarial data augmentation workflow.
We consider the problem of communication over a channel with a causal jamming adversary subject to quadratic constraints. A sender Alice wishes to communicate a message to a receiver Bob by transmitting a real-valued length-$n$ codeword $mathbf{x}=x_
1,...,x_n$ through a communication channel. Alice and Bob do not share common randomness. Knowing Alices encoding strategy, an adversarial jammer James chooses a real-valued length-n noise sequence $mathbf{s}=s_1,..,s_n$ in a causal manner, i.e., each $s_t (1<=t<=n)$ can only depend on $x_1,...,x_t$. Bob receives $mathbf{y}$, the sum of Alices transmission $mathbf{x}$ and James jamming vector $mathbf{s}$, and is required to reliably estimate Alices message from this sum. In addition, Alice and Jamess transmission powers are restricted by quadratic constraints $P>0$ and $N>0$. In this work, we characterize the channel capacity for such a channel as the limit superior of the optimal values of a series of optimizations. Upper and lower bounds on the optimal values are provided both analytically and numerically. Interestingly, unlike many communication problems, in this causal setting Alices optimal codebook may not have a uniform power allocation - for certain SNR, a codebook with a two-level uniform power allocation results in a strictly higher rate than a codebook with a uniform power allocation would.
In this work we consider the communication of information in the presence of an online adversarial jammer. In the setting under study, a sender wishes to communicate a message to a receiver by transmitting a codeword x=x_1,...,x_n symbol-by-symbol ov
er a communication channel. The adversarial jammer can view the transmitted symbols x_i one at a time, and can change up to a p-fraction of them. However, the decisions of the jammer must be made in an online or causal manner. More generally, for a delay parameter 0<d<1, we study the scenario in which the jammers decision on the corruption of x_i must depend solely on x_j for j < i - dn. In this work, we initiate the study of codes for online adversaries, and present a tight characterization of the amount of information one can transmit in both the 0-delay and, more generally, the d-delay online setting. We prove tight results for both additive and overwrite jammers when the transmitted symbols are assumed to be over a sufficiently large field F. Finally, we extend our results to a jam-or-listen online model, where the online adversary can either jam a symbol or eavesdrop on it. We again provide a tight characterization of the achievable rate for several variants of this model. The rate-regions we prove for each model are informational-theoretic in nature and hold for computationally unbounded adversaries. The rate regions are characterized by simple piecewise linear functions of p and d. The codes we construct to attain the optimal rate for each scenario are computationally efficient.
الأسئلة المقترحة
سجل دخول لتتمكن من نشر تعليقات
التعليقات
جاري جلب التعليقات
سجل دخول لتتمكن من متابعة معايير البحث التي قمت باختيارها
