Skip to main content

Table 3 Accuracy comparison among models (%)

From: Few-shot relation classification by context attention-based prototypical networks with BERT

Model5 way 5 shot5 way 10 shot10 way 5 shot10 way 10 shot
Meta network(CNN)80.03 ±0.5282.96 ± 0.5070.31 ± 0.4873.03 ± 0.44
GNN(CNN)77.75 ± 0.4480.56 ± 0.3866.02 ± 0.4069.30 ± 0.42
SNAIL(CNN)80.57 ± 0.2481.62 ± 0.2168.03 ± 0.2271.32± 0.20
Prototypical networks(CNN)85.57 ± 0.1488.17 ± 0.1075.01 ± 0.1678.50 ± 0.11
Proto_HATT(CNN)87.23 ± 0.0889.53 ± 0.0677.45 ± 0.0680.98 ± 0.08
Proto_CATT_BERT(CNN)94.86 ± 0.0495.74 ± 0.0590.01 ±0.0491.60 ± 0.03