Vulnerability Analysis of Chinese Digital Passwords Related to ATM PIN Using Deep Learning

Human-made digital passwords are a dominant form of ATM card authentication. However, it is vulnerable to guessing attacks. Unfortunately, previous studies have focused on letters-only passwords or passwords that include both letters and digits, and few empirical studies have been conducted on the security of digital passwords, let alone the regional differences in digital passwords. In this paper, we studied the vulnerability of pure six-digit passwords extracted from the leaked dataset of the Chinese website. We used the Pearson chi-square test to check whether each digit in the password obeyed a uniform distribution. We showed regional conventions for Chinese digital passwords. We proposed using a recurrent neural network (RNN) to model password resistance to guessing attacks and explore how different architectures and training methods impact the effectiveness of neural networks. The experimental results on five website datasets demonstrate the superior performance of the proposed approach over state-of-the-art deep learning techniques in terms of both learning efficiency and matching accuracy.