Blind Federated Learning without initial model

Journal of Big Data

Table 7 Blind FL results (accuracy-based weights)—Breast Cancer dataset

Agent	Size	% 1s	Accuracy	Accuracy	Precision	Precision
Agent	Size	% 1s	Pre-FL	Post-FL	Pre-FL	Post-FL
1	20%	14%	0.7727	0.9091	0.3750	0.6000
2	20%	26%	0.9565	0.9565	0.8571	0.8333
3	20%	35%	0.8696	0.8696	0.7778	0.7000
4	20%	13%	0.8261	0.8696	0.4286	0.5000
5	20%	26%	1.0000	1.0000	1.0000	0.8571
Avg.	–	–	0.8859	0.9209	0.6877	0.6981
1	19%	24%	0.9318	0.9545	0.8182	0.8889
2	26%	35%	0.8500	0.8667	0.8750	0.8824
3	16%	42%	0.8611	0.8889	0.9444	1.0000
4	15%	20%	0.9118	0.9412	0.7500	0.5833
5	24%	47%	0.8909	0.8909	0.9545	0.9500
Avg.	–	–	0.8891	0.9084	0.8684	0.8609
1	5%	27%	0.7500	0.7500	0.5000	0.5000
2	4%	43%	0.7000	0.8000	1.0000	1.0000
3	42%	35%	0.9375	0.9375	0.9375	0.8824
4	20%	23%	0.9778	0.9778	0.9091	0.9091
5	29%	49%	0.7273	0.7576	0.8571	0.9048
Avg.	–	–	0.8185	0.8446	0.8407	0.8392
1	3%	20%	0.8333	0.8333	0.5000	0.5000
2	4%	36%	0.8000	0.9000	1.0000	1.0000
3	45%	40%	0.9423	0.9423	0.9500	0.9487
4	21%	17%	0.9167	0.9167	0.7000	0.6364
5	26%	49%	0.8689	0.8689	1.0000	1.0000
Avg.	–	–	0.8722	0.8922	0.8300	0.8170