| Presentation | 2022-03-10 Experimental Evaluation of Influence of Distributing Deep Learning-Based IDSs on Their Classification Accuracy and Explainability Ayaka Oki, Yukio Ogawa, Kaoru Ota, Mianxiong Dong, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | Increased data traffic associated with the wide spread usage of IoT devices accentuates the risk of large-scale cyber attacks in the future. Intrusion detection systems (IDSs) thus need to be distributed in the edge computing for defending the attacks in parallel. The adoption of machine learning and eXplainable Artificial Intelligence (XAI) can improve the accuracy and reasoning estimation of IDSs, but the influence of the distribution on them are not clarified. We therefore simulate a distributed IDS and evaluate the influence on each attack category by decreasing the amount of training data given to the IDS. Our evaluations show that the accuracy decreases when the number of distributed IDSs is more than 100 and the precision also decreases by 10% to 30%. This is not only due to the lack of training data, but also the fact that the evidence features used for reasoning estimation have a higher similarity among different attack categories. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | distributed intrusion detection system / machine learning / explainable artificial intelligence |
| Paper # | IN2021-33 |
| Date of Issue | 2022-03-03 (IN) |
| Conference Information | |
| Committee | NS / IN |
|---|---|
| Conference Date | 2022/3/10(2days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | Online |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | General |
| Chair | Akihiro Nakao(Univ. of Tokyo) / Kenji Ishida(Hiroshima City Univ.) |
| Vice Chair | Tetsuya Oishi(NTT) / Kunio Hato(Internet Multifeed) |
| Secretary | Tetsuya Oishi(NTT) / Kunio Hato(Chuo Univ.) |
| Assistant | Kotaro Mihara(NTT) |
| Paper Information | |
| Registration To | Technical Committee on Network Systems / Technical Committee on Information Networks |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Experimental Evaluation of Influence of Distributing Deep Learning-Based IDSs on Their Classification Accuracy and Explainability |
| Sub Title (in English) | |
| Keyword(1) | distributed intrusion detection system |
| Keyword(2) | machine learning |
| Keyword(3) | explainable artificial intelligence |
| 1st Author's Name | Ayaka Oki |
| 1st Author's Affiliation | Muroran Institute of Technology(Muroran-IT) |
| 2nd Author's Name | Yukio Ogawa |
| 2nd Author's Affiliation | Muroran Institute of Technology(Muroran-IT) |
| 3rd Author's Name | Kaoru Ota |
| 3rd Author's Affiliation | Muroran Institute of Technology(Muroran-IT) |
| 4th Author's Name | Mianxiong Dong |
| 4th Author's Affiliation | Muroran Institute of Technology(Muroran-IT) |
| Date | 2022-03-10 |
| Paper # | IN2021-33 |
| Volume (vol) | vol.121 |
| Number (no) | IN-434 |
| Page | pp.pp.13-18(IN), |
| #Pages | 6 |
| Date of Issue | 2022-03-03 (IN) |