AIConfPaper
← Search

Dan Andrei Calian

4 accepted papers

2022

Defending Against Image Corruptions Through Adversarial Augmentations

Dan Andrei Calian, Florian Stimberg, Olivia Wiles, Sylvestre-Alvise Rebuffi, András György, Timothy A Mann +1
ICLR 2022poster

Modern neural networks excel at image classification, yet they remain vulnerable to common image corruptions such as blur, speckle noise or fog. Recent methods that focus on this problem, such as AugMix and DeepAugment, introduce defenses that operate in expectation over a distribution of image corr…

Cited by 54Source ↗PDF ↗Scholar ↗
2022

Hindering Adversarial Attacks with Implicit Neural Representations

Andrei A Rusu, Dan Andrei Calian, Sven Gowal, Raia Hadsell
ICML 2022spotlight

We introduce the Lossy Implicit Network Activation Coding (LINAC) defence, an input transformation which successfully hinders several common adversarial attacks on CIFAR-10 classifiers for perturbations up to 8/255 in Linf norm and 0.5 in L2 norm. Implicit neural representations are used to approxim…

Cited by 7Source ↗PDF ↗code ↗Scholar ↗
2021

Data Augmentation Can Improve Robustness

Sylvestre-Alvise Rebuffi, Sven Gowal, Dan Andrei Calian, Florian Stimberg, Olivia Wiles, Timothy Mann
NeurIPS 2021poster

Adversarial training suffers from robust overfitting, a phenomenon where the robust test accuracy starts to decrease during training. In this paper, we focus on reducing robust overfitting by using common data augmentation schemes. We demonstrate that, contrary to previous findings, when combined wi…

Cited by 410Source ↗PDF ↗code ↗Scholar ↗
2021

Improving Robustness using Generated Data

Sven Gowal, Sylvestre-Alvise Rebuffi, Olivia Wiles, Florian Stimberg, Dan Andrei Calian, Timothy Mann
NeurIPS 2021poster

Recent work argues that robust training requires substantially larger datasets than those required for standard classification. On CIFAR-10 and CIFAR-100, this translates into a sizable robust-accuracy gap between models trained solely on data from the original training set and those trained with ad…

Cited by 353Source ↗PDF ↗Scholar ↗