Physics-constrained offline RL pipeline for synthetic human mouse telemetry in behavioral-biometrics and continuous-authentication research (MuJoCo, d3rlpy, Balabit).

An open sourced approach to One-Shot Learning for Mouse Dynamics recognition in PyTorch. This includes tools for data preprocessing, training both classification and embedding models, and evaluating model performance on a Minecraft dataset.

Real-time AI stress tracker using mouse & keystroke dynamics. Privacy-first, local-only, and 100% offline.

Python project that uses machine learning to identify users based on their mouse movement patterns. It analyzes behavioral datasets and applies supervised learning techniques for behavior-based biometric authentication.