| 研究課題/領域番号 |
22KF0262
|
|---|
| 配分区分 | 基金 |
|---|
| 研究機関 | 奈良先端科学技術大学院大学 |
|---|
研究代表者 |
門林 雄基 奈良先端科学技術大学院大学, 先端科学技術研究科, 教授 (00294158)
|
|---|
| 研究分担者 |
BLUMBERGS BERNHARDS 奈良先端科学技術大学院大学, 先端科学技術研究科, 外国人特別研究員
|
|---|
| 研究期間 (年度) |
2023-03-08 – 2025-03-31
|
| キーワード | Situational awareness / Incident response / Threat intelligence / Distributed data mining |
|---|
| 研究実績の概要 |
Within the report period, main achievement is a successful prototype development, validation, and dataset collection. Complete prototype code and dataset are released publicly. It took unexpectedly significant time investment to research, develop, test, and validate the initial prototype as it is a novel concept and no existing related work has been identified. The work is described in a publication, which has been submitted and improved after receiving rejection from top-tier USENIX conference. The manuscript has been submitted to SECRYPT 2024 conference. Additionally, multiple invited presentations and guest lectures were given both domestically and internationally. As well as participating in conferences and community events to promote research and establish a professional network.
|
| 現在までの達成度 (区分) |
現在までの達成度 (区分)
2: おおむね順調に進展している
理由
Developing a novel approach based on the current cutting-edge technologies in data science, machine learning, cloud infrastructure engineering, and software engineering has its implicit challenges. Code development using newly developed libraries poses risks of limited functionality, operations not in line with documentation, and fixing the library code to improve its stability. All of these challenges are unavoidable in a situation, where an applied contribution is developed to be practically used by the incident response community. All of the risks so far have been addressed to permit delayed but steady progress in reaching the specified objectives.
|
| 今後の研究の推進方策 |
Currently, ongoing work is focused on collected data parsing, clustering, and pattern detection. The work should result in a research paper. Although the applicable machine learning and clustering algorithms have been well researched, problems may arise with correct data representation for these algorithms to function appropriately. This will come down to dataset engineering, model applicability, and evaluation. The issues may be tackled by improving the raw data collection, representation, and parsing approaches, as well as, consultations with data science and machine learning experts.
|
| 次年度使用額が生じた理由 |
It took unexpectedly significant time investment to research, develop, test, and validate the initial prototype as it is a novel concept and no existing related work has been identified. The work is described in a publication, which has been submitted and improved after receiving rejection from top-tier USENIX conference. The manuscript has been submitted to SECRYPT 2024 conference.
|
| 備考 |
投稿中の論文が出版されたのち、Webサイトを更新予定です。
|
