Skip to main content
Cornell University
We gratefully acknowledge support from the Simons Foundation, member institutions, and all contributors. Donate
arxiv logo > cs > arXiv:2106.02105v1

Help | Advanced Search

arXiv logo
Cornell University Logo

quick links

  • Login
  • Help Pages
  • About

Computer Science > Machine Learning

arXiv:2106.02105v1 (cs)
[Submitted on 3 Jun 2021 (this version), latest version 25 Oct 2021 (v2)]

Title:A Little Robustness Goes a Long Way: Leveraging Universal Features for Targeted Transfer Attacks

Authors:Jacob M. Springer, Melanie Mitchell, Garrett T. Kenyon
View a PDF of the paper titled A Little Robustness Goes a Long Way: Leveraging Universal Features for Targeted Transfer Attacks, by Jacob M. Springer and 2 other authors
View PDF
Abstract:Adversarial examples for neural network image classifiers are known to be transferable: examples optimized to be misclassified by a source classifier are often misclassified as well by classifiers with different architectures. However, targeted adversarial examples -- optimized to be classified as a chosen target class -- tend to be less transferable between architectures. While prior research on constructing transferable targeted attacks has focused on improving the optimization procedure, in this work we examine the role of the source classifier. Here, we show that training the source classifier to be "slightly robust" -- that is, robust to small-magnitude adversarial examples -- substantially improves the transferability of targeted attacks, even between architectures as different as convolutional neural networks and transformers. We argue that this result supports a non-intuitive hypothesis: on the spectrum from non-robust (standard) to highly robust classifiers, those that are only slightly robust exhibit the most universal features -- ones that tend to overlap with the features learned by other classifiers trained on the same dataset. The results we present provide insight into the nature of adversarial examples as well as the mechanisms underlying so-called "robust" classifiers.
Comments: 25 pages, 13 figures, 3 tables
Subjects: Machine Learning (cs.LG); Cryptography and Security (cs.CR)
Cite as: arXiv:2106.02105 [cs.LG]
  (or arXiv:2106.02105v1 [cs.LG] for this version)
  https://doi.org/10.48550/arXiv.2106.02105
arXiv-issued DOI via DataCite

Submission history

From: Jacob Springer [view email]
[v1] Thu, 3 Jun 2021 19:53:46 UTC (19,970 KB)
[v2] Mon, 25 Oct 2021 22:39:00 UTC (10,763 KB)
Full-text links:

Access Paper:

    View a PDF of the paper titled A Little Robustness Goes a Long Way: Leveraging Universal Features for Targeted Transfer Attacks, by Jacob M. Springer and 2 other authors
  • View PDF
  • Other Formats
license icon view license
Current browse context:
cs.LG
< prev   |   next >
new | recent | 2021-06
Change to browse by:
cs
cs.CR

References & Citations

  • NASA ADS
  • Google Scholar
  • Semantic Scholar

DBLP - CS Bibliography

listing | bibtex
Jacob M. Springer
Melanie Mitchell
Garrett T. Kenyon
a export BibTeX citation Loading...

BibTeX formatted citation

×
Data provided by:

Bookmark

BibSonomy logo Reddit logo

Bibliographic and Citation Tools

Bibliographic Explorer (What is the Explorer?)
Connected Papers (What is Connected Papers?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)

Code, Data and Media Associated with this Article

alphaXiv (What is alphaXiv?)
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Hugging Face (What is Huggingface?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)

Demos

Replicate (What is Replicate?)
Hugging Face Spaces (What is Spaces?)
TXYZ.AI (What is TXYZ.AI?)

Recommenders and Search Tools

Influence Flower (What are Influence Flowers?)
CORE Recommender (What is CORE?)
IArxiv Recommender (What is IArxiv?)
  • Author
  • Venue
  • Institution
  • Topic

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)
  • About
  • Help
  • contact arXivClick here to contact arXiv Contact
  • subscribe to arXiv mailingsClick here to subscribe Subscribe
  • Copyright
  • Privacy Policy
  • Web Accessibility Assistance
  • arXiv Operational Status
    Get status notifications via email or slack