arXiv research

AWP improves robustness by flattening weight loss landscape.

We introduce a new concept, data irrecoverability, and show that the well-studied concept of data privacy is sufficient but not necessary for data irrecoverability. We show that there are several regularized loss minimization problems that can use perturbed data with theoretical guarantees of generalization, i.e., loss…

This work improves structured prediction by learning the balance between signal and random noise.

EVILL uses randomised perturbations to improve exploration in bandit problems.

Neural networks are vulnerable to adversarial examples and researchers have proposed many heuristic attack and defense mechanisms. We address this problem through the principled lens of distributionally robust optimization, which guarantees performance under adversarial input perturbations. By considering a Lagrangian …

The scalability of submodular optimization methods is critical for their usability in practice. In this paper, we study the reducibility of submodular functions, a property that enables us to reduce the solution space of submodular optimization problems without performance loss. We introduce the concept of reducibility…

New algorithm reduces prediction errors across various loss functions.

Adversarial examples are intentionally perturbed images that mislead classifiers. These images can, however, be easily detected using denoising algorithms, when high-frequency spatial perturbations are used, or can be noticed by humans, when perturbations are large. In this paper, we propose EdgeFool, an adversarial im…

Study enhances robustness of In-CVaR based regression models under perturbation and contamination.

AMP regularization improves deep learning models by favoring flat minima.

Adversarial testing methods based on Projected Gradient Descent (PGD) are widely used for searching norm-bounded perturbations that cause the inputs of neural networks to be misclassified. This paper takes a deeper look at these methods and explains the effect of different hyperparameters (i.e., optimizer, step size an…

The paper analyzes and proposes methods for privately sharing individual privacy losses using per-instance differential privacy.

This paper analyzes adversarial attacks methods and their effectiveness.

We study the problem of online learning with non-convex losses, where the learner has access to an offline optimization oracle. We show that the classical Follow the Perturbed Leader (FTPL) algorithm achieves optimal regret rate of $O(T^{-1/2})$ in this setting. This improves upon the previous best-known regret rate of…

Adversarial weight perturbations can inject backdoors into trained neural models.

This research analyzes the consistency of convex and nonconvex surrogate losses for adversarially robust classification.

Margin-based structured prediction commonly uses a maximum loss over all possible structured outputs \cite{Altun03,Collins04b,Taskar03}. In natural language processing, recent work \cite{Zhang14,Zhang15} has proposed the use of the maximum loss over random structured outputs sampled independently from some proposal dis…

Paper introduces a new loss function for deep imbalanced classification.

A new method trains discrete EBMs without sampling.

Differentiable clustering method using perturbed spanning forests.

This paper improves SAM by reformulating it as a bilevel optimization problem.

MAP perturbation models have emerged as a powerful framework for inference in structured prediction. Such models provide a way to efficiently sample from the Gibbs distribution and facilitate predictions that are robust to random noise. In this paper, we propose a provably polynomial time randomized algorithm for learn…

SGD generalization bounds derived from information theory.

SapAugment learns adaptive augmentation policies for better model training.

MaxUp improves neural network training by minimizing worst case loss.

Paper introduces input perturbation for privacy in machine learning models.

SmoothDARTS stabilizes DARTS-based architecture search by smoothing loss landscapes.

Adversarial examples are perturbed inputs designed to fool machine learning models. Adversarial training injects such examples into training data to increase robustness. To scale this technique to large datasets, perturbations are crafted using fast single-step methods that maximize a linear approximation of the model'…

We study the problem of learning classifiers robust to universal adversarial perturbations. While prior work approaches this problem via robust optimization, adversarial training, or input transformation, we instead phrase it as a two-player zero-sum game. In this new formulation, both players simultaneously play the s…

Advances robustness of metric learning by adversarial margin in input space.

Optimal multiclass U-calibration error found to be Θ(√KT).

New method improves adversarial training robustness.

Neural Networks have been shown to be sensitive to common perturbations such as blur, Gaussian noise, rotations, etc. They are also vulnerable to some artificial malicious corruptions called adversarial examples. The adversarial examples study has recently become very popular and it sometimes even reduces the term "adv…

Simple regional perturbations maintain model transferability while reducing adversarial example distortion.

Adversarial examples are perturbed inputs designed to fool machine learning models. Most recent works on adversarial examples for image classification focus on directly modifying pixels with minor perturbations. A common requirement in all these works is that the malicious perturbations should be small enough (measured…

Paper develops robust Bayesian models for linear regression under adversarial perturbations.

Improved generalization bounds for SGD in non-convex learning.

Automates perturbation analysis for neural networks, enabling certified robustness on complex architectures.

Monge SAM improves deep learning by making sharpness-aware minimization invariant to reparametrizations.

Paper relaxes SGD privacy and generalization guarantees for non-smooth convex losses.

New algorithm controls systems with unknown, changing losses.

This paper characterizes and designs loss functions for robust classification with abstention.

Study privacy and accuracy in high-dimensional LASSO with perturbation mechanisms.

We consider the problem of controlling an unknown linear dynamical system in the presence of (nonstochastic) adversarial perturbations and adversarial convex loss functions. In contrast to classical control, the a priori determination of an optimal controller here is hindered by the latter's dependence on the yet unkno…

Machine learning models have been shown vulnerable to adversarial attacks launched by adversarial examples which are carefully crafted by attacker to defeat classifiers. Deep learning models cannot escape the attack either. Most of adversarial attack methods are focused on success rate or perturbations size, while we a…

New DP algorithm improves privacy and efficiency for convex optimization.

A new method generates natural-looking adversarial examples by bounding internal activation values.

Adversarial examples are perturbed inputs that are designed (from a deep learning network's (DLN) parameter gradients) to mislead the DLN during test time. Intuitively, constraining the dimensionality of inputs or parameters of a network reduces the 'space' in which adversarial examples exist. Guided by this intuition,…