Papers of the day   All papers

Adversarial Deepfakes: Evaluating Vulnerability of Deepfake Detectors to Adversarial Examples

Comments

roadrunner01: Adversarial Deepfakes: Evaluating Vulnerability of Deepfake Detectors to Adversarial Examples pdf: https://arxiv.org/pdf/2002.12749.pdf abs: https://arxiv.org/abs/2002.12749 project page: https://adversarialdeepfakes.github.io/ https://t.co/x97aGfL274

3 replies, 219 likes


Ariel Herbert-Voss: Very cool use of adversarial examples to break deepfake detection >:) There are not many applications of deep learning where adversarial examples present credible security risks - but this is one of the few applications where they do

3 replies, 87 likes


Paarth Neekhara: Excited to announce new work: Adversarial Deepfakes - CNN based Deepfake detectors that achieve state of the art results in detecting Deepfakes can be fooled using adversarial examples. Project page: https://adversarialdeepfakes.github.io/ Preprint: https://arxiv.org/abs/2002.12749 https://t.co/7rlIYYy9bg

2 replies, 48 likes


Károly Zsolnai-Fehér 📜: Sure, Deepfake Detectors Exist - But Can They Be Fooled? ▶️Full video (ours): https://youtu.be/hFZlxpJPI5w 📜Source paper: https://arxiv.org/abs/2002.12749 #ai #deeplearning #science #twominutepapers #deepfake #deepfakes https://t.co/De72GKkFfg

1 replies, 37 likes


Giorgio Patrini 🛡️👾: Nice work at the ⋂ of #deepfakes and adv examples: "Adversarial Deepfakes" https://arxiv.org/abs/2002.12749 Now empirically proven: one can craft adv examples for detectors. Unsurprising, but backed by extensive experiments. Also doable as black box attack & robust to video compression https://t.co/YTcoNBTSYX

0 replies, 21 likes


Sherif Hanna: Reason #25 why deepfake detection is a losing proposition in the long term and at scale.

0 replies, 10 likes


Nikos Sarris: Interesting research by Cornell University on deep fakes: Evaluating Vulnerability of Deepfake Detectors to Adversarial Examples https://arxiv.org/abs/2002.12749

0 replies, 7 likes


Shehzeen Hussain: Check out our latest work! https://twitter.com/n_paarth/status/1234567682487271425

1 replies, 2 likes


Big Wolf: Good job... When do we have a cost per fake analysis?? Inevitable. #AI #Robotics #smartcities #cybersecurity #tech #data #ar #vr #android #machinelearning #WearableTech #bitcoin #blockchain #deepfake #5g #automation #augmentation #facialrecognition #analysis

0 replies, 1 likes


Content

Found on Mar 02 2020 at https://arxiv.org/pdf/2002.12749.pdf

PDF content of a computer science paper: Adversarial Deepfakes: Evaluating Vulnerability of Deepfake Detectors to Adversarial Examples