Learning domain-heterogeneous speaker recognition systems with personalized continual federated learning

EURASIP Journal on Audio, Speech, and Music Processing

Table 5 Evaluation performance of C-PFL in continual learning settings

Evaluation groups	Canonical model	C-PFL Stage1	C-PFL Stage2	C-PFL Stage3
	EER	EER	EER	EER
G1	2.85	4.737	5.418	5.313
G2	33.06	13.844	13.828	13.844
G3	4.81	6.856	7.316	7.388
G4	39.23	19.751	19.158	18.523
G5	5.51	6.34	6.779	6.773
G6	37.80	16.605	16.876	15.695
G7	16.25	15.925	15.529	15.696
G8	41.29	24.111	23.69	22.913
G9	20.61	17.287	17.053	17.09
G10	37.61	24.512	23.69	22.913
G11	24.45	17.484	17.338	17.52
G12	37.10	22.649	22.434	22.117
mean(\(\mu\))	25.0475	14.965	14.922	14.678
std(\(\sigma\))	14.214	6.937	6.399	6.152