Table 12 Combined federated few-shot learning performance results.
Model component | Individual accuracy | Parameters | Training time | Memory usage |
|---|---|---|---|---|
RelationNetwork | 99.09% | 1,343,878 | Low | 0.4 GB |
MAML | 99.78% | 715,269 | Medium | 0.3 GB |
CrossTransformer | 93.59% | 25,794,566 | High | 1.2 GB |
FEAT | 99.76% | 4,397,829 | Medium | 0.6 GB |
Ensemble Fusion Results | ||||
Majority Vote | 99.9% | − | − | 2.5 GB |
Weighted Vote | 99.8% | − | − | 2.5 GB |
Confidence Weighted | 99.7% | − | − | 2.5 GB |
