Filters








12 Hits in 4.8 sec

Robustification of Segmentation Models Against Adversarial Perturbations In Medical Imaging [article]

Hanwool Park, Amirhossein Bayat, Mohammad Sabokrou, Jan S. Kirschke, Bjoern H. Menze
2020 arXiv   pre-print
This paper presents a novel yet efficient defense framework for segmentation models against adversarial attacks in medical imaging.  ...  In contrary to the defense methods against adversarial attacks for classification models which widely are investigated, such defense methods for segmentation models has been less explored.  ...  Conclusion We proposed the new defense methodology for defending the medical image semantic segmentation models against adversarial attacks.  ... 
arXiv:2009.11090v1 fatcat:noh6upann5fubich452qha7wza

Adversarial Attack and Defence through Adversarial Training and Feature Fusion for Diabetic Retinopathy Recognition

Sheeba Lal, Saeed Ur Rehman, Jamal Hussain Shah, Talha Meraj, Hafiz Tayyab Rauf, Robertas Damaševičius, Mazin Abed Mohammed, Karrar Hameed Abdulkareem
2021 Sensors  
We introduce a framework that provides a defensive model against the adversarial speckle-noise attack, the adversarial training, and a feature fusion strategy, which preserves the classification with correct  ...  Results obtained on the retinal fundus images, which are prone to adversarial attacks, are 99% accurate and prove that the proposed defensive model is robust.  ...  Defenses against Adversarial Attacks Reference [65] proposed defense against two groups: feature-level interpretation and model-level interpretation, input denoising, and model robustification.  ... 
doi:10.3390/s21113922 fatcat:ctlmaxj45bfdllzxclu7utc5we

Threat of Adversarial Attacks on Deep Learning in Computer Vision: A Survey [article]

Naveed Akhtar, Ajmal Mian
2018 arXiv   pre-print
For images, such perturbations are often too small to be perceptible, yet they completely fool the deep learning models.  ...  form of subtle perturbations to inputs that lead a model to predict incorrect outputs.  ...  To detect perturbation in a test image, its code is compared against those of training samples using the SVM.  ... 
arXiv:1801.00553v3 fatcat:xfk7togp5bhxvbxtwox3sckqq4

Adversarial Robustness in Multi-Task Learning: Promises and Illusions [article]

Salah Ghamizi, Maxime Cordy, Mike Papadakis, Yves Le Traon
2021 arXiv   pre-print
In particular, we show that the choice of the task to incorporate in the loss function are important factors that can be leveraged to yield more robust models.  ...  Vulnerability to adversarial attacks is a well-known weakness of Deep Neural networks.  ...  Answer to RQ1: For large perturbation budgets , l ∞ norms, or large models, multi-task learning does not reliably improve the robustness against adversarial attacks.  ... 
arXiv:2110.15053v1 fatcat:u4imbaiaszdeppeo73l7yvbzjm

A Review of Generative Adversarial Networks in Cancer Imaging: New Applications, New Solutions [article]

Richard Osuala, Kaisar Kushibar, Lidia Garrucho, Akis Linardos, Zuzanna Szafranowska, Stefan Klein, Ben Glocker, Oliver Diaz, Karim Lekadir
2021 arXiv   pre-print
The recent advancements in Generative Adversarial Networks (GANs) in computer vision as well as in medical imaging may provide a basis for enhanced capabilities in cancer detection and analysis.  ...  We analyse and discuss 163 papers that apply adversarial training techniques in the context of cancer imaging and elaborate their methodologies, advantages and limitations.  ...  To this end, we promote lines of research that use adversarial training schemes to target the robustification of segmentation models.  ... 
arXiv:2107.09543v1 fatcat:jz76zqklpvh67gmwnsdqzgq5he

JumpReLU: A Retrofit Defense Strategy for Adversarial Attacks [article]

N. Benjamin Erichson and Zhewei Yao and Michael W. Mahoney
2019 arXiv   pre-print
Our empirical results demonstrate that this increases model robustness, protecting against adversarial attacks with substantially increased levels of perturbations.  ...  In light of this, there has been a great deal of work on developing adversarial training strategies to improve model robustness.  ...  This can lead to problems in safety-and security-critical applications such as medical imaging, surveillance, autonomous driving, and voice command recognition.  ... 
arXiv:1904.03750v1 fatcat:w7jmyljosfb7bf3pvfbrm3lpgu

Statistical Methods with Applications in Data Mining: A Review of the Most Recent Works

Joaquim Fernando Pinto da Pinto da Costa, Manuel Cabral
2022 Mathematics  
The importance of statistical methods in finding patterns and trends in otherwise unstructured and complex large sets of data has grown over the past decade, as the amount of data produced keeps growing  ...  This paper is a comprehensive and systematic review of these recent developments in the area of data mining.  ...  Adversarial training (Goodfellow et al. 2015 ) is an attempt to solve this problem that has succeeded in improving generalization performance and made the model robust against adversarial perturbation  ... 
doi:10.3390/math10060993 fatcat:j5rz75qv6nburpq5rvsddw3cmu

Responsible and Regulatory Conform Machine Learning for Medicine: A Survey of Technical Challenges and Solutions [article]

Eike Petersen, Yannik Potdevin, Esfandiar Mohammadi, Stephan Zidowitz, Sabrina Breyer, Dirk Nowotka, Sandra Henn, Ludwig Pechmann, Martin Leucker, Philipp Rostalski, Christian Herzog
2021 arXiv   pre-print
wherever feasible, the use of inherently transparent models, comprehensive model testing and verification, as well as stakeholder inclusion.  ...  Machine learning is expected to fuel significant improvements in medical care.  ...  Conflict of interest statement EP and PR hold multiple patents with Dräger Medical. EM has received research funding by Volkswagen-Stiftung.  ... 
arXiv:2107.09546v1 fatcat:er3nlre7xrg4lmqsgxs7c4pswu

Responsible and Regulatory Conform Machine Learning for Medicine: A Survey of Challenges and Solutions

Eike Petersen, Yannik Potdevin, Esfandiar Mohammadi, Stephan Zidowitz, Sabrina Breyer, Dirk Nowotka, Sandra Henn, Ludwig Pechmann, Martin Leucker, Philipp Rostalski, Christian Herzog
2022 IEEE Access  
models, comprehensive out-of-distribution model testing and verification, as well as algorithmic impact assessments.  ...  We notice that distribution shift, spurious correlations, model underspecification, uncertainty quantification, and data scarcity represent severe challenges in the medical context.  ...  ., for assistance with the formatting of the references.  ... 
doi:10.1109/access.2022.3178382 fatcat:cwpkgkx2ibcgbdatd4aidwa4xy

Distributionally Robust Learning [article]

Ruidi Chen, Ioannis Ch. Paschalidis
2021 arXiv   pre-print
This monograph develops a comprehensive statistical learning framework that is robust to (distributional) perturbations in the data using Distributionally Robust Optimization (DRO) under the Wasserstein  ...  We consider a series of learning problems, including (i) distributionally robust linear regression; (ii) distributionally robust regression with group structure in the predictors; (iii) distributionally  ...  Particular mention is due to Bill Adams, at Boston Medical Center, whose efforts to make data available for research have been nothing short of extraordinary and who was instrumental in engaging the authors  ... 
arXiv:2108.08993v1 fatcat:6tsadkhvnrgwtk3etkvjumillq

Solving underdetermined inverse problems [article]

Maximilian Arthus März, Technische Universität Berlin, Martin Genzel, Pierre Weiss
2021
Such inverse problems arise in a wide range of applications, reaching from biomedical imaging modalities like computed tomography to seismic inversion in geophysics.  ...  Such schemes do not rely on an explicit formulation of a data model as in the first part, but infer structured sol [...]  ...  Our special thanks goes to Ali Hashemi, who referred us to Price's theorem, and Felix Voigtlaender, who assisted us by working out a fairly general version of it [85], which is also used in this work.  ... 
doi:10.14279/depositonce-12206 fatcat:bzdkctclcfaq5cfr756kw5efe4

Bridging Bayesian and Minimax Mean Square Error Estimation via Wasserstein Distributionally Robust Optimization [article]

Viet Anh Nguyen and Soroosh Shafieezadeh-Abadeh and Daniel Kuhn and Peyman Mohajerin Esfahani
2021 arXiv   pre-print
The proposed model can be viewed as a zero-sum game between a statistician choosing an estimator – that is, a measurable function of the observation – and a fictitious adversary choosing a prior – that  ...  We show that this algorithm enjoys a linear convergence rate and that its direction-finding subproblems can be solved in quasi-closed form.  ...  We are grateful to Erick Delage for valuable comments on an earlier version of this paper. This research was supported by the Swiss National Science Foundation grant number BSCGI0_157733.  ... 
arXiv:1911.03539v2 fatcat:dt2m4zfkjnfzjgobm32elznrga