Skip to main content

Robust Machine Learning Algorithms and Systems for Detection and Mitigation of Adversarial Attacks and Anomalies Proceedings of a Workshop (2019) / Chapter Skim
Currently Skimming:

Appendix D: Capability Technology Matrix
Pages 66-68

The Chapter Skim interface presents what we've algorithmically identified as the most significant single chunk of text within every page in the chapter.
Select key terms on the right to highlight them within pages of the chapter.


From page 66...
... Long-Term Capabilities (5-10 years) Matthew Turek,  Formalized requirements to test and assess Defense models Advanced  Verification of whether existing models can Research Projects run using new requirements Agency  Hybridized models that incorporate contextual information in addition to data points Hany Farid,  Creation of more automated processes and Dartmouth faster work, owing to the sheer volume of data College that is available  Improved accuracy  Use of secure-imaging pipelines to prevent the manipulation of digital evidence  Better cooperation from social media to reduce deepfakes  Increased education for citizen awareness to better recognize "fakes" in the digital age  Guidelines for responsible deployment of technologies  Consideration of the ethical and societal implications of algorithms that are advertised as artificial intelligence (AI)
Read the entire page →
From page 67...
... Terry Boult,  Open-set recognition algorithms for well-  Better network models for openUniversity of behaved, low-moderate dimensional feature set deep recognition Colorado, spaces  High-dimensional open-set Colorado Springs  Realistic large open-set data sets/protocols algorithms  Better understanding of image–feature relationships  Ability of iterative Layer-wise Origin-Target Synthesis (LOTS) to attack all kinds of systems  LOTS attacks are reasonably portable  Use of LOTS to build physical attacks/camouflage  Problems with good representations to relate images to features Rama Chellappa,  Explore the robustness of deeper networks  Keep changing the network University of  Work with multimodal inputs configuration and parameters in a Maryland,  Increase theoretical analysis probabilistic manner with College Park  Investigate how humans and machines can guaranteed performance (i.e., work together to thwart adversarial attacks adaptive networks)
Read the entire page →
From page 68...
...  Improve auto-calibration and understanding of when things have changed  Rethink the notion of domains in adaptation literature  Address the idea of overall robustness in the adaptation literature -- find a way to improve control over the initial model to reduce susceptibility to natural or artificial changes Anthony Hoogs,  Generative adversarial networks effectively  Human-level accuracy for action Kitware, Inc. applied to video and complex activity recognition  Cataloguing of large, common objects and in surveillance video events  Similar accuracy to humans but  Large graphical processing unit farms required much faster for video search and to keep up with video generation retrieval for complex activities  Structure learning of deep networks on a large and abstractions in Internet scale videos  Increased model transfer between video domains  Human-level accuracy for action recognition in single-action, temporarily clipped videos  Free-form, text-based queries with limited and open syntax and vocabulary  Adversarial attacks on video recognition problems (e.g., action recognition)
Read the entire page →

Key Terms

  • technology capabilities
  • adversarial attacks
  • easily adaptable
  • graphical processing
  • confidence intervals

  • difficult computer
  • principled evaluations
  • sheer volume
  • hybrid sensemaking
  • generative adversarial

  • universal semantic
  • complex activities
  • terry boult
  • machine massachusetts
  • faster work

  • adversarial networks
  • recognition colorado
  • probabilistic manner
  • machines work
  • capabilities described

  • domains requiring
  • attack real
  • realistic attacks
  • rama chellappa
  • information sciences

  • design networks
  • video search
  • increase theoretical
  • models easily
  • anish athalye


    • This material may be derived from roughly machine-read images, and so is provided only to facilitate research.
    More information on Chapter Skim is available.