Change search
ReferencesLink to record
Permanent link

Direct link
Methods for Increasing Robustness of Deep Convolutional Neural Networks
2015 (English)Independent thesis Advanced level (degree of Master (Two Years)), 80 credits / 120 HE creditsStudent thesis
Abstract [en]

Recent discoveries uncovered flaws in machine learning algorithms such as deep neural networks. Deep neural networks seem vulnerable to small amounts of non-random noise, created by exploiting the input to output mapping of the network. Applying this noise to an input image drastically decreases classication performance. Such image is referred to as an adversarial example. The purpose of this thesis is to examine how known regularization/robustness methods perform on adversarial examples. The robustness methods: dropout, low-pass filtering, denoising autoencoder, adversarial training and committees have been implemented, combined and tested. For the well-known benchmark, the MNIST (Mixed National Institute of Standards and Technology) dataset, the best combination of robustness methods has been found. Emerged from the results of the experiments, ensemble of models trained on adversarial examples is considered to be the best approach for MNIST. Harmfulness of the adversarial noise and some robustness experiments are demonstrated on CIFAR10 (The Canadian Institute for Advanced Research) dataset as well. Apart from robustness tests, the thesis describes experiments with human classification performance on noisy images and the comparison with performance of deep neural network.

Place, publisher, year, edition, pages
Keyword [en]
adversarial examples, deep neural network, noise robustness
National Category
Computer Science
URN: urn:nbn:se:hh:diva-29734OAI: diva2:867966
Subject / course
Computer science and engineering
Available from: 2015-11-11 Created: 2015-11-08 Last updated: 2015-11-16Bibliographically approved

Open Access in DiVA

fulltext(11449 kB)416 downloads
File information
File name FULLTEXT01.pdfFile size 11449 kBChecksum SHA-512
Type fulltextMimetype application/pdf

Computer Science

Search outside of DiVA

GoogleGoogle Scholar
Total: 416 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

Total: 648 hits
ReferencesLink to record
Permanent link

Direct link