| Presentation | 2024-03-22 Improved signature-embedding techniques against backdoor attacks on DNN models Akira Fujimoto, Yuntao Wang, Atsuko Miyaji, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | In recent years, machine learning, particularly deep learning, has made remarkable strides, and has great impact on our society across various domains such as transportation, healthcare, and finance. However, it is known that machine learning is highly vulnerable to malicious attacks. This paper focuses on the defense against backdoor attacks. A backdoor attack adds malicious data into the training dataset. The model trained on this dataset produces incorrect outputs for malicious data input by the attacker. A defense known as the signature-embedding method has been proposed. This defense involves incorporating data (signatures) that only the model creator adds into the training dataset to detect backdoor attacks. This paper highlights the problems with this defense method and proposes improvements. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | machine learning / deep neural network / backdoor attack |
| Paper # | ICSS2023-87 |
| Date of Issue | 2024-03-14 (ICSS) |
| Conference Information | |
| Committee | ICSS / IPSJ-SPT |
|---|---|
| Conference Date | 2024/3/21(2days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | OIST |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | Security, Trust, etc. |
| Chair | Daisuke Inoue(NICT) |
| Vice Chair | Akira Yamada(Kobe Univ.) / Toshihiro Yamauchi(Okayama Univ.) |
| Secretary | Akira Yamada(Mitsubishi Electric) / Toshihiro Yamauchi(Univ. of Electro-Comm.) |
| Assistant | Yo Kanemoto(NTT) / Masaya Sato(Okayama Prefectural Univ.) |
| Paper Information | |
| Registration To | Technical Committee on Information and Communication System Security / Special Interest Group on Security Psychology and Trust |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Improved signature-embedding techniques against backdoor attacks on DNN models |
| Sub Title (in English) | |
| Keyword(1) | machine learning |
| Keyword(2) | deep neural network |
| Keyword(3) | backdoor attack |
| 1st Author's Name | Akira Fujimoto |
| 1st Author's Affiliation | Osaka University(OU) |
| 2nd Author's Name | Yuntao Wang |
| 2nd Author's Affiliation | Osaka University(OU) |
| 3rd Author's Name | Atsuko Miyaji |
| 3rd Author's Affiliation | Osaka University(OU) |
| Date | 2024-03-22 |
| Paper # | ICSS2023-87 |
| Volume (vol) | vol.123 |
| Number (no) | ICSS-448 |
| Page | pp.pp.129-136(ICSS), |
| #Pages | 8 |
| Date of Issue | 2024-03-14 (ICSS) |