RePaLM: A Data-Driven AI Assistant for Making Stronger Pattern Choices.

Security mechanisms based on patterns, such as Pattern Lock, are commonly used to prevent unauthorized access. They introduce several benefits, such as ease of use, an additional layer of security, convenience, and versatility. However, many users tend to create simple and easily predictable patterns. To address this issue, we propose a data-driven real-time assistant approach called RePaLM. RePaLM is a neural network-based assistant that provides users with information about less commonly used pattern points, aiming to help users to make stronger, less predictable pattern choices. Our user study shows that RePaLM can effectively nudge users towards using less predictable patterns without compromising memorability. Overall, RePaLM is a promising solution for enhancing the security of pattern-based authentication systems.