Few-shot Image Classification on MiniImageNet 5-way 3-shot (test)
63.63AccuracyTraNFS-2
Evaluation results
| Method | Links | |
|---|---|---|
| TraNFS-2Noise Proportion=0%, Noise Type=outlier2022.04 | 63.63 | |
| TraNFS-3Noise Proportion=0%, Noise Type=outlier2022.04 | 63.61 | |
| Euclidean T = 50.0Method Type=Similarity weighted prototypes, Kernel=Euclidean, Temperature=50, Noise Proportion=0%, Noise Type=outlier2022.04 | 62.76 | |
| Cosine T = 2.0Method Type=Similarity weighted prototypes, Kernel=Cosine, Temperature=2, Noise Proportion=0%, Noise Type=outlier2022.04 | 62.72 | |
| Absolute T = 50.0Method Type=Similarity weighted prototypes, Kernel=Absolute, Temperature=50, Noise Proportion=0%, Noise Type=outlier2022.04 | 62.69 | |
| Cosine T = 5.0Method Type=Similarity weighted prototypes, Kernel=Cosine, Temperature=5, Noise Proportion=0%, Noise Type=outlier2022.04 | 62.63 | |
| OracleNoise Proportion=0%, Noise Type=outlier2022.04 | 62.6 | |
| Cosine T = 0.2Method Type=Similarity weighted prototypes, Kernel=Cosine, Temperature=0.2, Noise Proportion=0%, Noise Type=outlier2022.04 | 62.58 | |
| Euclidean T = 25.0Method Type=Similarity weighted prototypes, Kernel=Euclidean, Temperature=25, Noise Proportion=0%, Noise Type=outlier2022.04 | 62.57 | |
| Euclidean T = 10.0Method Type=Similarity weighted prototypes, Kernel=Euclidean, Temperature=10, Noise Proportion=0%, Noise Type=outlier2022.04 | 62.56 | |
| RNNPNoise Proportion=0%, Noise Type=outlier2022.04 | 62.55 | |
| MedianMethod Type=Spatial median prototypes, Noise Proportion=0%, Noise Type=outlier2022.04 | 62.53 | |
| Cosine T = 0.5Method Type=Similarity weighted prototypes, Kernel=Cosine, Temperature=0.5, Noise Proportion=0%, Noise Type=outlier2022.04 | 62.5 | |
| Cosine T = 1.0Method Type=Similarity weighted prototypes, Kernel=Cosine, Temperature=1, Noise Proportion=0%, Noise Type=outlier2022.04 | 62.5 | |
| Absolute T = 25.0Method Type=Similarity weighted prototypes, Kernel=Absolute, Temperature=25, Noise Proportion=0%, Noise Type=outlier2022.04 | 62.47 | |
| Vanilla ProtoNetNoise Proportion=0%, Noise Type=outlier2022.04 | 62.43 | |
| Linear ClassifierNoise Proportion=0%, Noise Type=outlier2022.04 | 61.58 | |
| Absolute T = 10.0Method Type=Similarity weighted prototypes, Kernel=Absolute, Temperature=10, Noise Proportion=0%, Noise Type=outlier2022.04 | 61.54 | |
| MAMLNoise Proportion=0%, Noise Type=outlier2022.04 | 59.76 | |
| Matching NetworksNoise Proportion=0%, Noise Type=outlier2022.04 | 57.82 | |
| OracleNoise Proportion=33.3%, Noise Type=outlier2022.04 | 56.89 | |
| TraNFS-2Noise Proportion=33.3%, Noise Type=outlier2022.04 | 54.75 | |
| TraNFS-3Noise Proportion=33.3%, Noise Type=outlier2022.04 | 54.72 | |
| Euclidean T = 10.0Method Type=Similarity weighted prototypes, Kernel=Euclidean, Temperature=10, Noise Proportion=33.3%, Noise Type=outlier2022.04 | 54.1 | |
| Absolute T = 25.0Method Type=Similarity weighted prototypes, Kernel=Absolute, Temperature=25, Noise Proportion=33.3%, Noise Type=outlier2022.04 | 54.07 | |
| MedianMethod Type=Spatial median prototypes, Noise Proportion=33.3%, Noise Type=outlier2022.04 | 53.82 | |
| Absolute T = 10.0Method Type=Similarity weighted prototypes, Kernel=Absolute, Temperature=10, Noise Proportion=33.3%, Noise Type=outlier2022.04 | 53.76 | |
| Absolute T = 50.0Method Type=Similarity weighted prototypes, Kernel=Absolute, Temperature=50, Noise Proportion=33.3%, Noise Type=outlier2022.04 | 53.73 | |
| Euclidean T = 25.0Method Type=Similarity weighted prototypes, Kernel=Euclidean, Temperature=25, Noise Proportion=33.3%, Noise Type=outlier2022.04 | 53.72 | |
| Euclidean T = 50.0Method Type=Similarity weighted prototypes, Kernel=Euclidean, Temperature=50, Noise Proportion=33.3%, Noise Type=outlier2022.04 | 53.55 | |
| Cosine T = 0.2Method Type=Similarity weighted prototypes, Kernel=Cosine, Temperature=0.2, Noise Proportion=33.3%, Noise Type=outlier2022.04 | 53.46 | |
| Cosine T = 5.0Method Type=Similarity weighted prototypes, Kernel=Cosine, Temperature=5, Noise Proportion=33.3%, Noise Type=outlier2022.04 | 53.19 | |
| Cosine T = 2.0Method Type=Similarity weighted prototypes, Kernel=Cosine, Temperature=2, Noise Proportion=33.3%, Noise Type=outlier2022.04 | 53.16 | |
| Nearest k = 1k=1, Noise Proportion=0%, Noise Type=outlier2022.04 | 53.07 | |
| Cosine T = 0.5Method Type=Similarity weighted prototypes, Kernel=Cosine, Temperature=0.5, Noise Proportion=33.3%, Noise Type=outlier2022.04 | 53.03 | |
| RNNPNoise Proportion=33.3%, Noise Type=outlier2022.04 | 52.88 | |
| Cosine T = 1.0Method Type=Similarity weighted prototypes, Kernel=Cosine, Temperature=1, Noise Proportion=33.3%, Noise Type=outlier2022.04 | 52.84 | |
| Vanilla ProtoNetNoise Proportion=33.3%, Noise Type=outlier2022.04 | 52.78 | |
| Linear ClassifierNoise Proportion=33.3%, Noise Type=outlier2022.04 | 51.21 | |
| Nearest k = 3k=3, Noise Proportion=0%, Noise Type=outlier2022.04 | 50.4 | |
| Nearest k = 5k=5, Noise Proportion=0%, Noise Type=outlier2022.04 | 50.24 | |
| Matching NetworksNoise Proportion=33.3%, Noise Type=outlier2022.04 | 48.56 | |
| MAMLNoise Proportion=33.3%, Noise Type=outlier2022.04 | 47.08 | |
| Nearest k = 1k=1, Noise Proportion=33.3%, Noise Type=outlier2022.04 | 44.66 | |
| Nearest k = 5k=5, Noise Proportion=33.3%, Noise Type=outlier2022.04 | 42.26 | |
| Nearest k = 3k=3, Noise Proportion=33.3%, Noise Type=outlier2022.04 | 41.59 |