Adversarial classification using signaling games with an application to phishing detection

详细信息    查看全文

• 作者：Nicolas Figueroa ; Gastón L’Huillier ; Richard Weber
• 关键词：Adversarial classification ; Signaling games ; Perfect Bayesian equilibrium ; Incremental learning ; Support vector machines ; Cybercrime ; Phishing filtering
• 刊名：Data Mining and Knowledge Discovery
• 出版年：2017
• 出版时间：January 2017
• 年：2017
• 卷：31
• 期：1
• 页码：92-133
• 全文大小：
• 刊物类别：Computer Science
• 刊物主题：Data Mining and Knowledge Discovery; Artificial Intelligence (incl. Robotics); Information Storage and Retrieval; Statistics for Engineering, Physics, Computer Science, Chemistry and Earth Sciences;
• 出版者：Springer US
• ISSN：1573-756X
• 卷排序：31

文摘

In adversarial classification, the interaction between classifiers and adversaries can be modeled as a game between two players. It is natural to model this interaction as a dynamic game of incomplete information, since the classifier does not know the exact intentions of the different types of adversaries (senders). For these games, equilibrium strategies can be approximated and used as input for classification models. In this paper we show how to model such interactions between players, as well as give directions on how to approximate their mixed strategies. We propose perceptron-like machine learning approximations as well as novel Adversary-Aware Online Support Vector Machines. Results in a real-world adversarial environment show that our approach is competitive with benchmark online learning algorithms, and provides important insights into the complex relations among players.