Overview
Study health at a glance.
Participants
–
Keystroke events
–
Task completions
–
across all rounds
ML dataset
–
Best model
No trained model yet — go to the Models tab.
Tasks collected by language / phase
| Participant | Created | Lang | Events (train/test) | Tasks | Status |
|---|
Train models on current data
Random Forest, KNN, SVM and MLP are trained on windowed keystroke features (dwell, flight, rollover, rhythm), evaluated by cross-validation.
Live identification demo
Type naturally in the box below. Keystroke timings are captured exactly as in the study, then the trained model predicts which participant's typing rhythm this most resembles — live. Windows are cut to the same geometry the model was trained on, so keep typing until the counter says ready. If no enrolled participant is a confident match, the result is reported as an unknown person rather than the nearest name.
0 keys captured
Polling every 3 s…
Typing right now
No new keystrokes since last poll.
Recently completed tasks
–
Newest sessions
–