Android-GAN: Defending against android pattern attacks using multi-modal generative network as anomaly detector

作者：

Highlights：

• GAN-based anomaly detection for discriminating real Android pattern among fake ones.

• Improved stability and performance of the network with replay buffer while training.

• Multi-modal system for touch trajectory and pressure outperforms uni-modal one.

• Extensive evaluation how pattern complexity and body posture affect the performance.

摘要

•GAN-based anomaly detection for discriminating real Android pattern among fake ones.•Improved stability and performance of the network with replay buffer while training.•Multi-modal system for touch trajectory and pressure outperforms uni-modal one.•Extensive evaluation how pattern complexity and body posture affect the performance.

论文关键词：Android pattern,GAN,LSTM,Anomaly detection,Authentication

论文评审过程：Received 24 March 2019, Revised 17 September 2019, Accepted 18 September 2019, Available online 19 September 2019, Version of Record 25 September 2019.

论文官网地址：https://doi.org/10.1016/j.eswa.2019.112964