A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf
.
Filters
Theoretically Principled Trade-off between Robustness and Accuracy
[article]
2019
arXiv
pre-print
We identify a trade-off between robustness and accuracy that serves as a guiding principle in the design of defenses against adversarial examples. ...
Inspired by our theoretical analysis, we also design a new defense method, TRADES, to trade adversarial robustness off against accuracy. ...
We thank Maria-Florina Balcan, Avrim Blum, Zico Kolter, and Aleksander Mądry for valuable comments and discussions. ...
arXiv:1901.08573v3
fatcat:tjkyieirgza5dfumq2mf2m4bpa
Towards Both Accurate and Robust Neural Networks without Extra Data
[article]
2022
arXiv
pre-print
robustness and accuracy. ...
Although incorporating extra training data can alleviate the trade-off to a certain extent, it remains unsolved to achieve both robustness and accuracy under limited training data. ...
Acknowledgements This work was partly supported by the National Key Research and Development Program of China (No. 2021ZD0200300) and the National Nature Science Foundation of China (No. 61836004). ...
arXiv:2103.13124v2
fatcat:w4px75xnfbaqvoai3sbq75uv2e
Can we have it all? On the Trade-off between Spatial and Adversarial Robustness of Neural Networks
[article]
2021
arXiv
pre-print
In this paper, we prove a quantitative trade-off between spatial and adversarial robustness in a simple statistical setting. ...
Towards achieving pareto-optimality in this trade-off, we propose a method based on curriculum learning that trains gradually on more difficult perturbations (both spatial and adversarial) to improve spatial ...
Acknowledgements and Funding Transparency Statement. ...
arXiv:2002.11318v5
fatcat:hisowgjwprg47nywdgme3vhwaa
BulletTrain: Accelerating Robust Neural Network Training via Boundary Example Mining
[article]
2021
arXiv
pre-print
reduction in clean and robust accuracy. ...
We apply our technique to several existing robust training algorithms and achieve a 2.1× speed-up for TRADES and MART on CIFAR-10 and a 1.7× speed-up for AugMix on CIFAR-10-C and CIFAR-100-C without any ...
On the other hand, NR controls the trade-off
between the robust accuracy and clean accuracy/theoretical speedup. ...
arXiv:2109.14707v2
fatcat:zlcqe2n3jzbzxjic7ycp3eh4vu
Adversarial Information Bottleneck
[article]
2021
arXiv
pre-print
curve achieve the best trade-off between compression and prediction, and has best robustness against various attacks. ...
How to optimize the IB principle for better robustness and figure out the effects of compression through the trade-off hyperparameter are two challenging problems. ...
Secondly, once the trade-off β exceeds the knee point, the clean accuracy and adversarial accuracy decease simultaneously. ...
arXiv:2103.00381v2
fatcat:7tsn47b7h5bu5o5ufo64am2zyq
Towards a theory of biological robustness
2007
Molecular Systems Biology
of Education, Sports, Culture, Science, and Technology). ...
Acknowledgements This research is supported, in part, by ERATO/SORST program (Japan Science and Technology Agency: JST), Sweden-Japan Strategic Collaboration Program (JST), Genome Network Project (Ministry ...
Does a trade-off between robustness and fragility indicate some kind of conservation principle as claimed by Csete and Doyle (2002) ? ...
doi:10.1038/msb4100179
pmid:17882156
pmcid:PMC2013924
fatcat:kwq2svrp5vhobhsuw7kudueyse
An Integrated Approach to Produce Robust Models with High Efficiency
[article]
2023
arXiv
pre-print
To solve the problems that adversarial training jeopardizes DNNs' accuracy on clean images and the struture of sparsity, we design a trade-off loss function that helps DNNs preserve their natural accuracy ...
Together with quantized EnResNet with trade-off loss function, we provide robust models that have high efficiency. ...
Models quantized by Binary-Relax have higher natural accuracy and adversarial accuracies than those quantized by Binary-Connect 4 Trade-off between robust accuracy and natural accuracy 4.1 Previous work ...
arXiv:2008.13305v4
fatcat:oqzevl6jwngotik5bclaf6onby
Accuracy Prevents Robustness in Perception-based Control
[article]
2020
arXiv
pre-print
In this paper we prove the existence of a fundamental trade-off between accuracy and robustness in perception-based control, where control decisions rely solely on data-driven, and often incompletely trained ...
Ultimately, our work proves the existence and the implications of a fundamental trade-off between accuracy and robustness in perception-based control, which, more generally, affects a large class of machine ...
In this paper, we characterize a fundamental trade-off between accuracy and robustness in a data-driven control problem. ...
arXiv:1910.00119v2
fatcat:23d2q7qyebgwxcz25asxhe5x5a
Design principles for biochemical oscillations with limited energy resources
[article]
2020
arXiv
pre-print
Here, we address the issue by deriving the energy accuracy and the sensitivity-accuracy trade-off relations for a general biochemical model, analytically and numerically. ...
understand the elementary principles for biochemical systems with limited energy resources to maintain phase accuracy and phase sensitivity. ...
Such trade-off relationship always holds when system parameters change, providing another design principle for biochemical systems to measure the balance between the sensory to external signals and the ...
arXiv:2004.00545v1
fatcat:knplm2vzpncvfmhgqv4euayt5m
Causality-Aided Trade-off Analysis for Machine Learning Fairness
[article]
2023
arXiv
pre-print
This paper uses causality analysis as a principled method for analyzing trade-offs between fairness parameters and other crucial metrics in ML pipelines. ...
Nonetheless, it is extremely difficult to analyze the trade-offs when there are multiple fairness parameters and other crucial metrics involved, coupled, and even in conflict with one another. ...
We also thank the participants in the human evaluation for their time and effort. We acknowledge Timothy Menzies for his insightful suggestions regarding the quality of learned causal graphs. ...
arXiv:2305.13057v3
fatcat:ssiyqfropzepfprj5wnvesaxdq
Adversarial Training Reduces Information and Improves Transferability
2021
PROCEEDINGS OF THE THIRTIETH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE AND THE TWENTY-EIGHTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE
We show that the Adversarial Training can improve linear transferability to new tasks, from which arises a new trade-off between transferability of representations and accuracy on the source task. ...
We validate our results employing robust networks trained on CIFAR-10, CIFAR-100 and ImageNet on several datasets. ...
Theoretically principled tradeoff between robustness and accuracy. arXiv preprint arXiv:1901.08573 . Figure 1 : 1 Figure 1: Inversion using the standard and variational ResNet-50 model. ...
doi:10.1609/aaai.v35i3.16371
fatcat:ef7i47mvhvan5nyii6smzilgja
COVID-19 Imaging Data Privacy by Federated Learning Design: A Theoretical Framework
[article]
2020
arXiv
pre-print
with scalability and robustness. ...
To address COVID-19 healthcare challenges, we need frequent sharing of health data, knowledge and resources at a global scale. ...
One dimension differentiates between privacy and utility and highlights their trade-off. ...
arXiv:2010.06177v1
fatcat:asfm6ub2krc7bj3p2vyugqfdiu
On the uncertainty principle of neural networks
[article]
2022
arXiv
pre-print
Various empirical and analytic studies have substantiated that there is more or less a trade-off between the accuracy and robustness of neural networks. ...
To more deeply explore and understand this issue, in this study we show that the accuracy-robustness trade-off is an intrinsic property whose underlying mechanism is closely related to the uncertainty ...
there is more or less a trade-off between the accuracy and the robustness of a general neural network. ...
arXiv:2205.01493v3
fatcat:ksxouwwckrdrbhhkfxhnd2aim4
Do Wider Neural Networks Really Help Adversarial Robustness?
[article]
2021
arXiv
pre-print
Specifically, we show that the model robustness is closely related to the tradeoff between natural accuracy and perturbation stability, which is controlled by the robust regularization parameter λ. ...
However, it remains elusive how neural network width affects model robustness. In this paper, we carefully examine the relationship between network width and model robustness. ...
Zhang et al. (2019) theoretically studied the trade-off between natural accuracy and robust accuracy for adversarially trained models. ...
arXiv:2010.01279v3
fatcat:4cnm2pz6avaezckfasd4kid3fi
A Closer Look at Accuracy vs. Robustness
[article]
2020
arXiv
pre-print
between robustness and accuracy. ...
We conclude that achieving robustness and accuracy in practice may require using methods that impose local Lipschitzness and augmenting them with deep learning generalization techniques. ...
Acknowledgments and Disclosure of Funding Kamalika Chaudhuri and Yao-Yuan Yang thank NSF under CIF 1719133, CNS 1804829 and IIS 1617157 for support. ...
arXiv:2003.02460v3
fatcat:tl5xijlxy5hahhgfsxzqjtkkvy
« Previous
Showing results 1 — 15 out of 68,968 results