Neutral mention · rule
Yes, these are “adversarial patches” in image classification, like https://arxiv.org/abs/1712.09665 . Similarly you can take these adversaries and add them to your own larger model, in an arms race.
Back to the collection
Adversarial Patch
Tom B. Brown, Dandelion Mané, Aurko Roy, Martín Abadi, Justin Gilmer
2017 · arXiv
- HN score
- 0
- Mentions
- 5
- Readers
- 5
- Months
- 5
From the abstract
We present a method to create universal, robust, targeted adversarial image patches in the real world. The patches are universal because they can be used to attack any scene, robust because they work under a wide variety of transformations, and targeted because they can cause a classifier to output any target class. These adversarial patches can be printed, added to any scene, photographed, and presented to image classifiers; even when the patches are small, they cause the classifiers to ignore the other items in the scene and report a chosen target class. To reproduce the results from the paper, our code is available at https://github.com/tensorflow/cleverhans/tree/master/examples/adversarial_patch
Metadata: arXiv. Abstracts retain their original authors’ rights.
In the conversation
Original HN comments, with labels from simple text rules. A label describes the whole comment and may miss context or irony. Read the discussion to make up your own mind. About the labels
5 selected excerpts · 5 total mentions
Neutral mention · rule
Honestly, I'd wear that dress easily, even for regular outings, not just for hiding from cameras. Those hoodies, though, maybe I'm an old fart but they're a bit too "abstract". Sadly, those seem like the most effective option right now. I read the Adversarial Patch article [0] and it pretty much demonstrated similar patterns. Trouble is, if these get widespread enough - they won't look out of place but the moment they become widespread is the moment updates get pushed to make them obsolete. It's an unescapable situation. [0]: https://arxiv.org/abs/1712.09665
Neutral mention · rule
The article mentions the use of "Juggalo" style face paint to avoid face detection. This type of countermeasure is effective against conventional HOG-SVM type face detectors such as [1], but would be inadvisable for daily use. To deal with more advanced convolutional detectors, consider wearing a hat or clothing with adversarial patches [2], simple patterns that can be printed onto stickers. These patches work by feeding the detector inputs that cause an overwhelming bias toward a particular class (eg. toaster), which drowns out the real signal (eg. human face). Even the first generation of adversarial patches are effective against most detectors deployed as of 2020. [1] http://dlib.net/face_landmark_detection.py.html [2] Brown et al. https://arxiv.org/abs/1712.09665
Neutral mention · rule
For better or worse, this helps establish neural networks that are not (or less?) vulnerable to deception via e.g. fake eyeglasses [ LINK: https://dl.acm.org/citation.cfm?id=2978392 ] and adversarial stickers [ LINK: https://arxiv.org/abs/1712.09665 ]. From the post: "We’ve developed a method to assess whether a neural network classifier can reliably defend against adversarial attacks not seen during training [such as Elastic, Fog, Gabor, and Snow]. Our method yields a new metric, UAR (Unforeseen Attack Robustness), which evaluates the robustness of a single model against an unanticipated attack, and highlights the need to measure performance across a more diverse range…
Read the full comment
of unforeseen attacks." Paper: https://arxiv.org/abs/1908.08016 Code: https://github.com/ddkang/advex-uar If anyone w/relevant expertise is willing to share thoughts on this: please do. For further reading: this just in -> "Robust Machine Learning Algorithms and Systems for Detection and Mitigation of Adversarial Attacks and Anomalies: Proceedings of a Workshop" (2019) an open-access publication from the National Academies of Sciences, Engineering, and Medicine: https://doi.org/10.17226/25534.
Neutral mention · rule
Masks, face altering, stickers, etc. I like this research about "Adversarial Patches" - stickers that confuse image recognition systems. I imagine one may wear a display with ever changing images that are salient and confuse such systems. > Research Article: https://arxiv.org/pdf/1712.09665.pdf > Video for patches: https://www.youtube.com/watch?v=i1sp4X57TL4
Keep following the thread.
More in Machine learningMore from the same topics.
2020arXiv
Language Models are Few-Shot Learners
Tom B. Brown, Benjamin Mann, Nick Ryder +28
Recent work has demonstrated substantial gains on many NLP tasks and benchmarks by pre-training on a large corpus of text followed by fine-tuning on a specific task.
83 mentions35 months
3HN score
2016arXiv
Understanding deep learning requires rethinking generalization
Chiyuan Zhang, Samy Bengio, Moritz Hardt +2
Despite their massive size, successful deep artificial neural networks can exhibit a remarkably small difference between training and test performance.
16 mentions14 months
2.5HN score
2016arXiv
End to End Learning for Self-Driving Cars
Mariusz Bojarski, Davide Del Testa, Daniel Dworakowski +10
We trained a convolutional neural network (CNN) to map raw pixels from a single front-facing camera directly to steering commands.
8 mentions7 months
2.5HN score
2018arXiv
Categorizing Variants of Goodhart's Law
David Manheim, Scott Garrabrant
There are several distinct failure modes for overoptimization of systems on the basis of metrics.
8 mentions7 months
2.5HN score