Table 4 Clustering results on the three data sets with three clustering methods. Significant values are in [bold and italic].
From: Transductive zero-shot learning via knowledge graph and graph convolutional networks
AWA2 | ImageNet50 | ImageNet100 | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
Seen | Unseen | H | Seen | Unseen | H | Seen | Unseen | H | ||
GCNZ | GloVe300 | 67.19 | 33.34 | 44.57 | 95.33 | 20.92 | 34.31 | 85.37 | 16.06 | 27.03 |
SGCN | GloVe300 | 94.03 | 36.74 | 52.84 | 96.80 | 25.38 | 40.22 | 92.18 | 17.30 | 29.13 |
DGP | GloVe300 | 94.20 | 36.78 | 52.90 | 96.76 | 23.91 | 38.34 | 92.15 | 15.95 | 27.19 |
TGCNZ_1 | GloVe300+SCAN | 66.23 | 39.98 | 49.86 | 95.39 | 24.58 | 39.09 | 84.56 | 18.42 | 30.25 |
TGCNZ_2 | GloVe300+SCAN | 64.59 | 41.19 | 50.75 | 94.39 | 26.87 | 41.83 | 85.89 | 18.72 | 30.74 |
TGCNZ_1 | GloVe300+GAT-Cluster | 65.09 | 43.90 | 52.44 | 92.39 | 24.20 | 38.35 | 85.64 | 16.77 | 28.05 |
TGCNZ_2 | GloVe300+GAT-Cluster | 66.67 | 47.36 | 55.38 | 92.20 | 26.02 | 40.59 | 84.96 | 17.19 | 28.59 |
TGCNZ_1 | GloVe300+K-means | 67.00 | 33.88 | 45.00 | 93.97 | 23.57 | 37.69 | 85.36 | 17.43 | 28.95 |
TGCNZ_2 | GloVe300+K-means | 67.88 | 34.54 | 45.78 | 93.89 | 24.90 | 39.36 | 84.91 | 17.58 | 29.13 |
