Using Supervised Learning to Strengthen Password Security

Abstract

There's no doubt that text-based passwords will continue to dominate the authentication market. Machine learning and deep learning algorithms may assist developers in evaluating their strengths and predicting their susceptibility to brute-force assaults, since these passwords usually consist of meaningful strings. Long Short-Term Memory (LSTM) and Generative Adversarial Networks (GAN) are two examples of advanced algorithms that may learn user habits and utilize that information to construct lists of anticipated and similar text passwords. In this research, we investigate the feasibility of classifying passwords as either strong, moderate, or weak using machine learning methods. We also assess the potential of deep learning and machine learning to understand the patterns used by hashing algorithms. Additionally, we have created a model for password creation that utilizes Gated Recurrent Unit (GRU) to generate new passwords according to patterns that have been learnt. We want to make password creation and management easier for users and increase password security using this method. Topics—guessing passwords, GRU, LSTM, GAN, RNN.