Blind Federated Learning without initial model

Journal of Big Data

Table 14 Blended Blind FL results (Precision-based weights, slope 5, sigmoid)—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%	0.8696	0.9130	0.7143	0.8333
3	20%	35%	0.9565	0.9565	1.0000	0.7778
4	20%	13%	0.9130	0.9130	0.6000	0.5000
5	20%	26%	0.8696	0.9565	0.6667	0.8571
Avg.	–	–	0.8763	0.9296	0.6712	0.7137
1	19%	24%	0.8750	0.9063	0.6250	0.6667
2	26%	35%	0.9524	0.9524	0.9583	0.8519
3	16%	42%	0.7073	0.7317	0.9231	1.0000
4	15%	20%	0.9231	0.9231	0.7500	0.7500
5	24%	47%	0.8704	0.8889	0.8696	0.9091
Avg.	–	–	0.8656	0.8805	0.8252	0.8355
1	5%	27%	0.8462	0.8462	0.6667	0.6667
2	4%	43%	1.0000	1.0000	1.0000	1.0000
3	42%	35%	0.8969	0.8969	0.8684	0.8684
4	20%	23%	0.8974	0.9231	0.7500	0.8182
5	29%	49%	0.8986	0.8986	1.0000	1.0000
Avg.	–	–	0.9078	0.9129	0.8570	0.8707
1	3%	20%	0.8750	0.8750	0.6667	0.4000
2	4%	36%	0.8462	0.8462	1.0000	1.0000
3	45%	40%	0.8889	0.8990	0.8421	0.8421
4	21%	17%	0.9024	0.9024	0.5000	0.3636
5	26%	49%	0.8971	0.9118	1.0000	1.0000
Avg.	–	–	0.8819	0.8869	0.8018	0.7211