Enhancing Adversarial Defense in Behavioral Authentication Systems Through Random Projections

Md Islam, Md Anam

2024

Abstract

Behavioral Authentication (BA) systems employ a verification algorithm to verify users based on their behavior patterns. To eliminate the need for a profile database to store the profiles and to enhance the system’s performance, the verification algorithm usually trains a Neural Network (NN) classifier on user profiles. However, like other NN applications, the NN-based BA classifiers are also susceptible to adversarial attacks. To defend against such attacks, we employed a method that adds noise to the training data by using Random Projection (RP) and its reverse process. This approach prevents model overfitting and maintains the model’s predictions at an expected level. Our technique has also proven effective against attacks based on adversarial examples. We tested our proposed method on two BA systems, achieving the expected classification accuracy. Furthermore, the attacks based on adversarial examples are significantly less effective against BA classifiers trained with noisy data compared to those trained with plain data. Our approach is general and can be applied to other BA systems to protect their classifiers from similar attacks.

Paper Citation

in Harvard Style

Islam M. and Anam M. (2024). Enhancing Adversarial Defense in Behavioral Authentication Systems Through Random Projections. In Proceedings of the 21st International Conference on Security and Cryptography - Volume 1: SECRYPT; ISBN 978-989-758-709-2, SciTePress, pages 758-763. DOI: 10.5220/0012792800003767

in Bibtex Style

@conference{secrypt24,
author={Md Islam and Md Anam},
title={Enhancing Adversarial Defense in Behavioral Authentication Systems Through Random Projections},
booktitle={Proceedings of the 21st International Conference on Security and Cryptography - Volume 1: SECRYPT},
year={2024},
pages={758-763},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0012792800003767},
isbn={978-989-758-709-2},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 21st International Conference on Security and Cryptography - Volume 1: SECRYPT
TI - Enhancing Adversarial Defense in Behavioral Authentication Systems Through Random Projections
SN - 978-989-758-709-2
AU - Islam M.
AU - Anam M.
PY - 2024
SP - 758
EP - 763
DO - 10.5220/0012792800003767
PB - SciTePress