Keystroke Prediction

Keystroke prediction research focuses on leveraging patterns in typing behavior for various applications, including user authentication, academic integrity monitoring, and security vulnerability assessment. Current research explores diverse model architectures, such as transformers and recurrent neural networks, often combined with innovative feature engineering techniques like distance-based metrics or multi-sensor data fusion (e.g., acoustic and accelerometer data from headphones). These advancements have implications for enhancing security systems, improving online proctoring, and understanding human-computer interaction, but also raise concerns about privacy vulnerabilities from side-channel attacks.