We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:

References & Citations

DBLP - CS Bibliography


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Machine Learning

Title: Improving Network Robustness against Adversarial Attacks with Compact Convolution

Abstract: Though Convolutional Neural Networks (CNNs) have surpassed human-level performance on tasks such as object classification and face verification, they can easily be fooled by adversarial attacks. These attacks add a small perturbation to the input image that causes the network to misclassify the sample. In this paper, we focus on neutralizing adversarial attacks by compact feature learning. In particular, we show that learning features in a closed and bounded space improves the robustness of the network. We explore the effect of L2-Softmax Loss, that enforces compactness in the learned features, thus resulting in enhanced robustness to adversarial perturbations. Additionally, we propose compact convolution, a novel method of convolution that when incorporated in conventional CNNs improves their robustness. Compact convolution ensures feature compactness at every layer such that they are bounded and close to each other. Extensive experiments show that Compact Convolutional Networks (CCNs) neutralize multiple types of attacks, and perform better than existing methods in defending adversarial attacks, without incurring any additional training overhead compared to CNNs.
Subjects: Machine Learning (cs.LG); Cryptography and Security (cs.CR); Machine Learning (stat.ML)
Cite as: arXiv:1712.00699 [cs.LG]
  (or arXiv:1712.00699v2 [cs.LG] for this version)

Submission history

From: Rajeev Ranjan [view email]
[v1] Sun, 3 Dec 2017 03:09:31 GMT (1426kb,D)
[v2] Thu, 22 Mar 2018 21:45:44 GMT (2117kb,D)

Link back to: arXiv, form interface, contact.