arXiv research

The paper proposes effective margin regularization to improve adversarial robustness in deep neural networks.

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

For linear classifiers, the relationship between (normalized) output margin and generalization is captured in a clear and simple bound -- a large output margin implies good generalization. Unfortunately, for deep models, this relationship is less clear: existing analyses of the output margin give complicated bounds whi…

This work improves adversarial robustness by boosting model ensembles with margin maximization.

Adversarial robustness has become an important research topic given empirical demonstrations on the lack of robustness of deep neural networks. Unfortunately, recent theoretical results suggest that adversarial training induces a strict tradeoff between classification accuracy and adversarial robustness. In this paper,…

We study adversarial robustness of neural networks from a margin maximization perspective, where margins are defined as the distances from inputs to a classifier's decision boundary. Our study shows that maximizing margins can be achieved by minimizing the adversarial loss on the decision boundary at the "shortest succ…

Efficiently computes robust option prices using multi-marginal martingale transport.

How can we make machine learning provably robust against adversarial examples in a scalable way? Since certified defense methods, which ensure $ε$-robust, consume huge resources, they can only achieve small degree of robustness in practice. Lipschitz margin training (LMT) is a scalable certified defense, but it can als…

Deep models, while being extremely versatile and accurate, are vulnerable to adversarial attacks: slight perturbations that are imperceptible to humans can completely flip the prediction of deep models. Many attack and defense mechanisms have been proposed, although a satisfying solution still largely remains elusive. …

CMRM improves robustness in noisy label settings without requiring privileged knowledge.

Improved robustness of machine learning models with controlled Lipschitz constants.

Relief based algorithms have often been claimed to uncover feature interactions. However, it is still unclear whether and how interaction terms will be differentiated from marginal effects. In this paper, we propose IMMIGRATE algorithm by including and training weights for interaction terms. Besides applying the large …

Proposes SOVR loss to improve adversarial robustness by increasing logit margins.

Data augmentation (DA) is commonly used during model training, as it significantly improves test error and model robustness. DA artificially expands the training set by applying random noise, rotations, crops, or even adversarial perturbations to the input data. Although DA is widely used, its capacity to provably impr…

Adversarial training is a technique for training robust machine learning models. To encourage robustness, it iteratively computes adversarial examples for the model, and then re-trains on these examples via some update rule. This work analyzes the performance of adversarial training on linearly separable data, and prov…

Interpolation hurts robust generalization even without noise.

Improved exploration in RL with latent state marginalization.

Robust GP model detects and corrects sparse outliers.

In this work, we study a new approach to optimizing the margin distribution realized by binary classifiers. The classical approach to this problem is simply maximization of the expected margin, while more recent proposals consider simultaneous variance control and proxy objectives based on robust location estimates, in…

Theory for soft-margin classifiers on object manifolds.

Proposes RLAR for efficient labeled data classification with robust margin and manifold structure.

New method improves calibration of neural networks by targeting robust margins and local smoothness.

Extends martingale transport for robust finance problems.

Enhances robustness of deep neural networks with randomized smoothing.

We present a new boosting algorithm, motivated by the large margins theory for boosting. We give experimental evidence that the new algorithm is significantly more robust against label noise than existing boosting algorithm.

Develops a new robustness criterion for VAEs and provides theoretical guarantees.

The paper improves robust optimization by introducing margin theory.

In this paper, we provide some results on Skorokhod embedding with local time and its applications to the robust hedging problem in finance. First we investigate the robust hedging of options depending on the local time by using the recently introduced stochastic control approach, in order to identify the optimal hedgi…

Study improves adversarial classification using distributionally robust models.

Study bounds financial path expectations using martingale distributions.

AdaBoost improves binary classification in robust one-bit compressed sensing with adversarial errors.

MARVEL curbs memorization of noisy labels in deep nets.

Proposes MDR estimator for unbiased OPE with large action spaces.

This work improves adversarial robustness in sparse coding models.

We propose the Margin Adaptation for Generative Adversarial Networks (MAGANs) algorithm, a novel training procedure for GANs to improve stability and performance by using an adaptive hinge loss function. We estimate the appropriate hinge loss margin with the expected energy of the target distribution, and derive princi…

In this paper, we study the implicit regularization of the gradient descent algorithm in homogeneous neural networks, including fully-connected and convolutional neural networks with ReLU or LeakyReLU activations. In particular, we study the gradient descent or gradient flow (i.e., gradient descent with infinitesimal s…

New findings on robust learning with well-separated data.

Adversarial training is by far the most successful strategy for improving robustness of neural networks to adversarial attacks. Despite its success as a defense mechanism, adversarial training fails to generalize well to unperturbed test set. We hypothesize that this poor generalization is a consequence of adversarial …

This work analyzes the maximum-margin bias in quasi-homogeneous neural networks.

Jointly learns feature and sample relevancies for robust sparse recovery.

New framework for estimating treatment effects in observational studies.

New algorithm reduces regret in GLM bandits with tighter bounds.

In this article we discuss the problem of calculating optimal model-independent (robust) bounds for the price of Asian options with discrete and continuous averaging. We will give geometric characterisations of the maximising and the minimising pricing model for certain types of Asian options in discrete and continuous…

Study finds a method to discover causal relationships that are invariant to marginal distributions.

This paper solves robust utility maximization with unknown claim dependencies.

MLS improves feature selection for imbalanced data.

New estimator reduces variance in off-policy evaluation for contextual bandits.

MSBM extends SB for multi-marginal trajectory inference.