Presentation 2020-05-09
On Power and limitation of adversarial example attacks
Kouichi Sakurai,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) A risk of adversarial example attacks which cause deep learning to make wrong decisions is getting serious even from a cybersecurity perspective. Research papers on various attacks and defenses have already been published. Some studies have claimed that the adversarial attack are a kind of an intrinsic property of deep learning itself. A famous crypto analysist Adi Shamir introduced combinatorial geometry theory to elucidate the adversarial example attacks and experimentally succeeded to confirm the validity of his attack search algorithm. In this article, in addition to an introduction of Shamir's research, we report recent research progress on the power and limitations of adversarial example attacks.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Deep learning / dversarial examples / activate function / combinatoric geometry / cyber science
Paper # COMP2020-5
Date of Issue 2020-05-02 (COMP)

Conference Information
Committee COMP / IPSJ-AL
Conference Date 2020/5/9(1days)
Place (in Japanese) (See Japanese page)
Place (in English) National Institute of Informatics
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair Toshihiro Fujito(Toyohashi Univ. of Tech.)
Vice Chair Shinichi Nakano(Gunma Univ.)
Secretary Shinichi Nakano(Nagoya Univ) / (Univ. of Hyogo)
Assistant Kazuhisa Seto(Seikei Univ.)

Paper Information
Registration To Technical Committee on Theoretical Foundations of Computing / Special Interest Group on Algorithms
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) On Power and limitation of adversarial example attacks
Sub Title (in English)
Keyword(1) Deep learning
Keyword(2) dversarial examples
Keyword(3) activate function
Keyword(4) combinatoric geometry
Keyword(5) cyber science
1st Author's Name Kouichi Sakurai
1st Author's Affiliation Kyushu University(Kyushu Univ.)
Date 2020-05-09
Paper # COMP2020-5
Volume (vol) vol.120
Number (no) COMP-13
Page pp.pp.33-36(COMP),
#Pages 4
Date of Issue 2020-05-02 (COMP)