Detecting Fraudulent Behavior Using Recurrent Neural Networks

• Item type: Symposium(1)
• 公開日: 2016-10-04
• タイトル: Detecting Fraudulent Behavior Using Recurrent Neural Networks
• タイトル / 言語: en
• タイトル: Detecting Fraudulent Behavior Using Recurrent Neural Networks
• 言語: eng
• キーワード / 主題Scheme: Other
• キーワード / 主題: Fraud Detection，Machine Learning，Recurrent Neural Network
• 資源タイプ / 資源タイプ識別子: http://purl.org/coar/resource_type/c_5794
• 資源タイプ: conference paper
• 著者所属: Institute of Information Security/Yahoo Japan Corporation
• 著者所属: Yahoo Japan Corporation
• 著者所属: Institute of Information Security
• 著者所属(英): en
• 著者所属(英): Institute of Information Security / Yahoo Japan Corporation
• 著者所属(英): en
• 著者所属(英): Yahoo Japan Corporation
• 著者所属(英): en
• 著者所属(英): Institute of Information Security
• 著者名: Yoshihiro, Ando; Hidehito, Gomi; Hidehiko, Tanaka
• 著者名(英): Yoshihiro, Ando; Hidehito, Gomi; Hidehiko, Tanaka
• 論文抄録(英) / 内容記述タイプ: Other
• 論文抄録(英) / 内容記述: Due to an increase in illegal accesses to Internet services, detecting and preventing them has become important. To prevent an illegal access, not only rule-based techniques but also machine learning techniques have been used. In the field of security, such as malware detection, machine learning techniques are used to learn behavioral patterns and detect those that are suspicious.However, there are few studies targeting the behavioral patterns of a malicious user engaged in fraudulent acts on Internet services. In this paper, we propose the approach which uses the patterns in web access logs to detect fraudulent behaviors. We apply and evaluate a recurrent neural network (RNN) to recognize these behaviors. Our result indicates that RNN is very effective for fraudulent behavior detection.
• 書誌レコードID / 識別子タイプ: NCID
• 書誌レコードID / 関連識別子: ISSN 1882-0840
• 書誌情報: コンピュータセキュリティシンポジウム2016論文集; 巻 2016, 号 2, p. 805-810, 発行日 2016-10-04
• 出版者 / 言語: ja
• 出版者: 情報処理学会