Robust image classification: analysis and applications

Alhussein Fawzi · 2016

Summary

Novel quantitative results that precisely describe the behavior of classifiers under perturbations of the data are provided, including a blessing of dimensionality phenomenon: in high-dimensional classification tasks, robustness to random noise can be achieved, even if the classifier is extremely unstable to adversarial perturbation.

Published: 2016-01-01
Cited by: 1
References: 134
DOI: https://doi.org/10.5075/epfl-thesis-7258
Semantic Scholar: https://api.semanticscholar.org/CorpusID:126032745

References

Learning in the Limit with Adversarial DisturbancesConstantine Caramanis, Shie Mannor — 2008
Deep learning via Hessian-free optimizationJames Martens — 2010
Manifolds and Differential GeometryJeffrey Lee — 2009
Sur le problème de la divisionStanisław Łojasiewicz — 1959
Manitest: Are classifiers really invariant?Alhussein Fawzi, Pascal Frossard — 2015
Signal recovery from Pooling RepresentationsJoan Bruna Estrach, Arthur Szlam, Yann LeCun — 2013
Seeking Categories in the BrainSimon J. Thorpe, Michèle Fabre‐Thorpe — 2001
On Invariance and Selectivity in Representation LearningFabio Anselmi, Lorenzo Rosasco, Tomaso Poggio — 2015
Intriguing properties of neural networksChristian Szegedy, Wojciech Zaremba, Ilya Sutskever — 2013
Analysis of classifiers’ robustness to adversarial perturbationsAlhussein Fawzi, Omar Fawzi, Pascal Frossard — 2015
Very Deep Convolutional Networks for Large-Scale Image RecognitionKaren Simonyan, Andrew Zisserman — 2014
The Caltech-UCSD Birds-200-2011 DatasetCatherine Wah — 2011
How Can Deep Rectifier Networks Achieve Linear Separability and Preserve Distances?Senjian An, Farid Boussaïd, Mohammed Bennamoun — 2015
Towards Deep Neural Network Architectures Robust to Adversarial ExamplesShixiang Gu, Luca Rigazio — 2014
Domain-size pooling in local descriptors: DSP-SIFTJingming Dong, Stefano Soatto — 2014
Understanding deep image representations by inverting themAravindh Mahendran, Andrea Vedaldi — 2014
Multi-view feature engineering and learningJingming Dong, Nikolaos Karianakis, Damek Davis — 2015
Deep roto-translation scattering for object classificationEdouard Oyallon, Stéphane Mallat — 2014
MatConvNet: Convolutional Neural Networks for MATLABAndrea Vedaldi, Karel Lenc — 2014
Robust support vector machines for classification and computational issuesTheodore B. Trafali̇s, Robin C. Gilbert — 2007

Cited by

Learning embeddings: efficient algorithms and applicationsC. Jose — 2018

Related papers

No related papers recorded.