Classification on AV-MNIST
72.38AccuracyI2M2
Evaluation Results
| Method | Links | |
|---|---|---|
| I2M2Base Fusion strategy=LRTF2024.05 | 72.38 | |
| I2M2Base Fusion strategy=LF2024.05 | 72.34 | |
| LFModality=Inter-modality, Fusion strategy=Late Fusion2024.05 | 71.68 | |
| LRTFModality=Inter-modality, Fusion strategy=Low-rank multimodal fusion2024.05 | 71.49 | |
| ConFuModality=Audio-Visual fusion, Evaluation Protocol=Zero-shot2025.11 | 71.2 | |
| SMIL (BL) + RegBNModality=Multimodal, Norm. params. (#)=02023.10 | 71.1 | |
| SON-GOKU + AdaTask2025.09 | 71 | |
| SymileModality=Audio-Visual fusion, Evaluation Protocol=Zero-shot2025.11 | 70.9 | |
| SMIL (BL) + PMDNModality=Multimodal, Norm. params. (#)=11,1362023.10 | 70.7 | |
| SMIL (BL)Modality=Multimodal, Norm. params. (#)=02023.10 | 70.6 | |
| FAMO2025.09 | 70 | |
| SON-GOKU + PCGrad2025.09 | 70 | |
| SON-GOKUvariant=Quantile, Density-Matched=true2025.09 | 69 | |
| SON-GOKU + GradNorm2025.09 | 69 | |
| SON-GOKU2025.09 | 69 | |
| INTRA-MODALITYModality=Intra-modality2024.05 | 68.63 | |
| SON-GOKUvariant=Threshold, Density-Matched=true2025.09 | 68 | |
| SON-GOKUvariant=Quantile2025.09 | 68 | |
| AdaTask2025.09 | 67 | |
| FairGrad2025.09 | 66 | |
| Nash-MTL2025.09 | 66 | |
| Static One-Shot2025.09 | 66 | |
| SON-GOKUvariant=kNN-Symm., Density-Matched=true2025.09 | 66 | |
| GradNorm2025.09 | 65 | |
| PCGrad2025.09 | 65 | |
| SON-GOKUvariant=kNN-Symm.2025.09 | 65 | |
| TRIANGLEModality=Audio-Visual fusion, Evaluation Protocol=Zero-shot2025.11 | 64.9 | |
| IMAGE-ONLYModality=Image only2024.05 | 64.73 | |
| ConFuModality=Vision, Evaluation Protocol=Zero-shot2025.11 | 64.6 | |
| GRAMModality=Audio-Visual fusion, Evaluation Protocol=Zero-shot2025.11 | 64.4 | |
| SMIL (BL)Modality=Unimodal (image), Norm. params. (#)=02023.10 | 64.3 | |
| SON-GOKUvariant=Signed-only, Density-Matched=true2025.09 | 64 | |
| GRAMModality=Vision, Evaluation Protocol=Zero-shot2025.11 | 63.9 | |
| CLIPModality=Vision, Evaluation Protocol=Zero-shot2025.11 | 63 | |
| Uniform2025.09 | 63 | |
| SON-GOKUvariant=Signed-only2025.09 | 63 | |
| Tri-CLIPModality=Vision, Evaluation Protocol=Zero-shot2025.11 | 62.5 | |
| MGDA2025.09 | 62 | |
| CAGrad2025.09 | 62 | |
| Single-Step2025.09 | 59 | |
| SMIL (BL)Modality=Unimodal (audio), Norm. params. (#)=02023.10 | 43.2 | |
| CLIPModality=Audio, Evaluation Protocol=Zero-shot2025.11 | 41.1 | |
| Tri-CLIPModality=Audio, Evaluation Protocol=Zero-shot2025.11 | 40 | |
| ConFuModality=Audio, Evaluation Protocol=Zero-shot2025.11 | 39.7 | |
| AUDIO-ONLYModality=Audio only2024.05 | 39.59 | |
| GRAMModality=Audio, Evaluation Protocol=Zero-shot2025.11 | 9.8 |