Table 3. Person identification accuracies in the MPIIDPEye dataset using differentially private eye movement features with majority voting.

	Person identification accuracies (k-NN|SVM|DT|RF)
Method	ϵ = 0.48	ϵ = 2.4	ϵ = 4.8	ϵ = 24	ϵ = 48
FPA	1	1	1	1	1
CFPA-32	0.15|0.08|0.44|0.37	0.13|0.08|0.46|0.39	0.11|0.08|0.48|0.41	0.40|0.11|0.64|0.70	0.72|0.16|0.87|0.92
CFPA-64	0.14|0.08|0.42|0.34	0.13|0.08|0.44|0.37	0.12|0.08|0.45|0.38	0.39|0.11|0.63|0.71	0.70|0.17|0.85|0.92
CFPA-128	0.16|0.05|0.39|0.36	0.15|0.05|0.41|0.36	0.17|0.05|0.43|0.39	0.45|0.07|0.55|0.63	0.70|0.16|0.74|0.88
DCFPA-32	0.06|0.10|0.39|0.37	0.06|0.10|0.39|0.36	0.08|0.10|0.39|0.36	0.10|0.10|0.39|0.37	0.10|0.10|0.40|0.38
DCFPA-64	0.10|0.10|0.33|0.35	0.10|0.10|0.32|0.34	0.10|0.10|0.32|0.33	0.13|0.10|0.31|0.34	0.13|0.10|0.32|0.33
DCFPA-128	0.09|0.05|0.24|0.28	0.09|0.05|0.25|0.27	0.10|0.05|0.23|0.27	0.10|0.06|0.24|0.26	0.10|0.05|0.22|0.25