Blind Federated Learning without initial model

Journal of Big Data

Table 18 Blind FL results (constant weights, slope 5, sigmoid)—adult dataset

Agent	Size	% 1s	Accuracy	Accuracy	Precision	Precision
Agent	Size	% 1s	Pre-FL	Post-FL	Pre-FL	Post-FL
1	20%	72%	0.9444	0.9722	0.9310	0.9310
2	20%	75%	0.8333	0.9722	0.8286	0.9667
3	20%	75%	0.8889	0.9167	0.8667	0.8966
4	20%	67%	0.9444	0.9722	0.9655	0.8788
5	20%	72%	0.9444	0.9722	1.0000	0.9333
Avg.	–	–	0.9111	0.9611	0.9184	0.9213
1	13%	63%	0.3227	0.3642	0.8571	0.9063
2	30%	80%	0.3008	0.3384	0.7500	1.0000
3	21%	87%	0.3324	0.3526	0.9048	0.9355
4	11%	61%	0.3701	0.4156	0.9091	0.9444
5	24%	81%	0.3479	0.3673	0.9398	0.9091
Avg.	–	–	0.3348	0.3676	0.8722	0.9391
1	4%	75%	0.2561	0.3049	0.7500	0.8333
2	5%	68%	0.4030	0.4328	1.0000	1.0000
3	45%	77%	0.4079	0.4207	0.9241	0.9200
4	13%	51%	0.4840	0.5479	0.5385	0.7333
5	32%	85%	0.2759	0.2969	0.9143	0.8889
Avg.	–	–	0.3654	0.4006	0.8254	0.8751
1	1%	25%	0.6563	0.6875	0.5000	1.0000
2	5%	72%	0.2857	0.3000	0.6667	0.7500
3	45%	78%	0.3320	0.3561	0.8319	0.8780
4	14%	53%	0.5539	0.5637	0.9189	0.8750
5	35%	89%	0.2994	0.3307	0.8961	0.9756
Avg.	–	–	0.4255	0.4476	0.7627	0.8957