An investigation of privacy preservation in deep learning-based eye-tracking

Table 2 Performance metrics for SVM in validation and test data with both instance and person label

	Valid_instance	Valid_person	Test_instance	Test_person
Accuracy	0.85	0.81	0.79	0.77
F1-score	0.82	0.80	0.80	0.79
AUROC	0.92	0.90	0.88	0.86
AP (average precision)	0.92	0.91	0.87	0.91

Threshold have been adjusted to achieve the best performance on the validation set in each model (0.68 for instance and 0.8 for person model).

ISSN: 1475-925X