Multimodal Classification on CUB multimodal noise η=10%, ε=5 original (test)
72.4AccuracyTAHCD
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| TAHCDTest Noise Level=eta=10%, epsilon=5, Evaluation Protocol=Train on original / Test on corrupted2026.01 | 72.4 | 72.4 | 71.8 | |
| PDFTest Noise Level=eta=10%, epsilon=5, Evaluation Protocol=Train on original / Test on corrupted2026.01 | 62.2 | 62.3 | 62.4 | |
| QMFTest Noise Level=eta=10%, epsilon=5, Evaluation Protocol=Train on original / Test on corrupted2026.01 | 61.5 | 61.3 | 61.1 | |
| MDTest Noise Level=eta=10%, epsilon=5, Evaluation Protocol=Train on original / Test on corrupted2026.01 | 57.4 | 57.7 | 57.6 | |
| MLCLNetTest Noise Level=eta=10%, epsilon=5, Evaluation Protocol=Train on original / Test on corrupted2026.01 | 54.8 | 54.9 | 54.9 | |
| DisNCLTest Noise Level=eta=10%, epsilon=5, Evaluation Protocol=Train on original / Test on corrupted2026.01 | 52.2 | 52.2 | 52.1 | |
| SPSTest Noise Level=eta=10%, epsilon=5, Evaluation Protocol=Train on original / Test on corrupted2026.01 | 52.1 | 52.1 | 52.7 | |
| ASLTest Noise Level=eta=10%, epsilon=5, Evaluation Protocol=Train on original / Test on corrupted2026.01 | 51.5 | 51.4 | 51.2 | |
| SMILETest Noise Level=eta=10%, epsilon=5, Evaluation Protocol=Train on original / Test on corrupted2026.01 | 49.2 | 49.1 | 49.1 |