Image Classification on TinyImageNet (val)
90.65AccuracyMbLS
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| MbLSModel=ViT2021.11 | 90.65 | 1.26 | — | — | — | — | — | — | |
| LSModel=ViT2021.11 | 90.5 | 2.37 | — | — | — | — | — | — | |
| FLSDModel=ViT2021.11 | 90.47 | 4.25 | — | — | — | — | — | — | |
| FLModel=ViT2021.11 | 90.39 | 4.51 | — | — | — | — | — | — | |
| ALIAS Adam versionWeight decay (wd)=true, Momentum parameter (beta)=true, Cosine schedule (cosine sc)=true, Backbone (Swin)=Swin Transformer2025.06 | 79.161 | — | — | — | — | — | — | — | |
| SIGN-SGDWeight decay (wd)=true, Momentum parameter (beta)=true, Learning rate (lr)=true, Cosine schedule (cosine sc)=true, Backbone (Swin)=Swin Transformer2025.06 | 78.885 | — | — | — | — | — | — | — | |
| NORMALIZED SGDWeight decay (wd)=true, Momentum parameter (beta)=true, Learning rate (lr)=true, Cosine schedule (cosine sc)=true, Backbone (Swin)=Swin Transformer2025.06 | 78.375 | — | — | — | — | — | — | — | |
| PRODIGYWeight decay (wd)=true, Momentum parameter (beta)=true, Cosine schedule (cosine sc)=true, Backbone (Swin)=Swin Transformer2025.06 | 77.944 | — | — | — | — | — | — | — | |
| ADAMWWeight decay (wd)=true, Momentum parameter (beta)=true, Learning rate (lr)=true, Cosine schedule (cosine sc)=true, Backbone (Swin)=Swin Transformer2025.06 | 77.612 | — | — | — | — | — | — | — | |
| STEEPEST DESCNETWeight decay (wd)=true, Momentum parameter (beta)=true, Learning rate (lr)=true, Cosine schedule (cosine sc)=true, Backbone (Swin)=Swin Transformer2025.06 | 77.547 | — | — | — | — | — | — | — | |
| ALIAS Adam versionWeight decay (wd)=true, Momentum parameter (beta)=true, Cosine schedule (cosine sc)=false, Backbone (Swin)=Swin Transformer2025.06 | 77.433 | — | — | — | — | — | — | — | |
| SIGN-SGDWeight decay (wd)=true, Momentum parameter (beta)=true, Learning rate (lr)=true, Cosine schedule (cosine sc)=false, Backbone (Swin)=Swin Transformer2025.06 | 77.045 | — | — | — | — | — | — | — | |
| PRODIGYWeight decay (wd)=true, Momentum parameter (beta)=true, Cosine schedule (cosine sc)=false, Backbone (Swin)=Swin Transformer2025.06 | 77.035 | — | — | — | — | — | — | — | |
| DYRNoise rate=0.0, Selection=Best2023.02 | 65.14 | — | — | — | — | — | — | — | |
| DYRNoise rate=0.0, Epoch selection=Best2023.02 | 65.14 | — | — | — | — | — | — | — | |
| Proposed-LMNoise rate=0.0, Selection=Best2023.02 | 64.58 | — | — | — | — | — | — | — | |
| Proposed-LMNoise rate=0.0, Epoch selection=Best2023.02 | 64.58 | — | — | — | — | — | — | — | |
| Proposed-MNoise rate=0.0, Epoch selection=Best2023.02 | 64.37 | — | — | — | — | — | — | — | |
| Proposed-LMNoise rate=0.0, Selection=Last2023.02 | 64.15 | — | — | — | — | — | — | — | |
| Proposed-LMNoise rate=0.0, Epoch selection=Last2023.02 | 64.15 | — | — | — | — | — | — | — | |
| Proposed-LNoise rate=0.0, Selection=Best2023.02 | 64.07 | — | — | — | — | — | — | — | |
| Proposed-LNoise rate=0.0, Epoch selection=Best2023.02 | 64.07 | — | — | — | — | — | — | — | |
| Proposed-MNoise rate=0.0, Epoch selection=Last2023.02 | 64.01 | — | — | — | — | — | — | — | |
| DYRNoise rate=0.0, Selection=Last2023.02 | 63.97 | — | — | — | — | — | — | — | |
| DYRNoise rate=0.0, Epoch selection=Last2023.02 | 63.97 | — | — | — | — | — | — | — | |
| DE-CENoise rate=0.0, Selection=Best2023.02 | 63.62 | — | — | — | — | — | — | — | |
| DE-CENoise rate=0.0, Epoch selection=Best2023.02 | 63.62 | — | — | — | — | — | — | — | |
| Proposed-LNoise rate=0.0, Selection=Last2023.02 | 63.21 | — | — | — | — | — | — | — | |
| Proposed-LNoise rate=0.0, Epoch selection=Last2023.02 | 63.21 | — | — | — | — | — | — | — | |
| M-DYRNoise rate=0.0, Selection=Best2023.02 | 62.32 | — | — | — | — | — | — | — | |
| M-DYRNoise rate=0.0, Epoch selection=Best2023.02 | 62.32 | — | — | — | — | — | — | — | |
| DE-CENoise rate=0.0, Selection=Last2023.02 | 61.28 | — | — | — | — | — | — | — | |
| DE-CENoise rate=0.0, Epoch selection=Last2023.02 | 61.28 | — | — | — | — | — | — | — | |
| Single-CENoise rate=0.0, Epoch selection=Best2023.02 | 61.19 | — | — | — | — | — | — | — | |
| Proposed-MNoise rate=0.3x, Epoch selection=Best2023.02 | 61.06 | — | — | — | — | — | — | — | |
| M-DYRNoise rate=0.0, Selection=Last2023.02 | 61.04 | — | — | — | — | — | — | — | |
| M-DYRNoise rate=0.0, Epoch selection=Last2023.02 | 61.04 | — | — | — | — | — | — | — | |
| Proposed-LMNoise rate=0.3x, Epoch selection=Best2023.02 | 61.01 | — | — | — | — | — | — | — | |
| FGAModel=VGG-19, Attack Type=BadNets, Poison Rate=10%2026.03 | 60.84 | — | — | — | 99.86 | 89.96 | — | — | |
| FGAModel=VGG-19, Attack Type=Blend, Poison Rate=10%2026.03 | 60.84 | — | — | — | 80.98 | 99.06 | — | — | |
| FGAModel=VGG-19, Attack Type=WaNet, Poison Rate=10%2026.03 | 60.84 | — | — | — | 75.5 | 89.38 | — | — | |
| FGAModel=VGG-19, Attack Type=Input-Aware, Poison Rate=10%2026.03 | 60.84 | — | — | — | 94.08 | 70.82 | — | — | |
| DYRNoise rate=0.3x, Epoch selection=Best2023.02 | 60.74 | — | — | — | — | — | — | — | |
| Proposed-LMNoise rate=0.1, Selection=Best2023.02 | 60.73 | — | — | — | — | — | — | — | |
| M-DYRNoise rate=0.3x, Epoch selection=Best2023.02 | 60.7 | — | — | — | — | — | — | — | |
| Proposed-LNoise rate=0.3x, Epoch selection=Best2023.02 | 60.68 | — | — | — | — | — | — | — | |
| Proposed-MNoise rate=0.4x, Epoch selection=Best2023.02 | 60.51 | — | — | — | — | — | — | — | |
| M-DYRNoise rate=0.1, Selection=Best2023.02 | 60.4 | — | — | — | — | — | — | — | |
| Single-CENoise rate=0.0, Epoch selection=Last2023.02 | 60.26 | — | — | — | — | — | — | — | |
| Proposed-MNoise rate=0.3x, Epoch selection=Last2023.02 | 60.26 | — | — | — | — | — | — | — | |
| Proposed-LNoise rate=0.1, Selection=Best2023.02 | 60.25 | — | — | — | — | — | — | — | |
| Proposed-LMNoise rate=0.3x, Epoch selection=Last2023.02 | 60.23 | — | — | — | — | — | — | — | |
| Proposed-LMNoise rate=0.4x, Epoch selection=Best2023.02 | 60.12 | — | — | — | — | — | — | — | |
| DYRNoise rate=0.1, Selection=Best2023.02 | 60.04 | — | — | — | — | — | — | — | |
| DE-CENoise rate=0.1, Selection=Best2023.02 | 60.03 | — | — | — | — | — | — | — | |
| DE-CENoise rate=0.3x, Epoch selection=Best2023.02 | 60.03 | — | — | — | — | — | — | — | |
| Proposed-LNoise rate=0.3x, Epoch selection=Last2023.02 | 59.99 | — | — | — | — | — | — | — | |
| Proposed-LMNoise rate=0.1, Selection=Last2023.02 | 59.95 | — | — | — | — | — | — | — | |
| DE-CENoise rate=0.4x, Epoch selection=Best2023.02 | 59.94 | — | — | — | — | — | — | — | |
| CORES^2Noise rate=0.0, Selection=Best2023.02 | 59.74 | — | — | — | — | — | — | — | |
| CORES^2Noise rate=0.0, Epoch selection=Best2023.02 | 59.74 | — | — | — | — | — | — | — | |
| M-DYRNoise rate=0.2, Selection=Best2023.02 | 59.59 | — | — | — | — | — | — | — | |
| Proposed-LNoise rate=0.4x, Epoch selection=Best2023.02 | 59.54 | — | — | — | — | — | — | — | |
| DE-CENoise rate=0.3x, Epoch selection=Last2023.02 | 59.51 | — | — | — | — | — | — | — | |
| M-DYRNoise rate=0.4x, Epoch selection=Best2023.02 | 59.45 | — | — | — | — | — | — | — | |
| Proposed-MNoise rate=0.4x, Epoch selection=Last2023.02 | 59.34 | — | — | — | — | — | — | — | |
| Proposed-LNoise rate=0.1, Selection=Last2023.02 | 59.31 | — | — | — | — | — | — | — | |
| DYRNoise rate=0.3x, Epoch selection=Last2023.02 | 59.25 | — | — | — | — | — | — | — | |
| DYRNoise rate=0.4x, Epoch selection=Best2023.02 | 59.2 | — | — | — | — | — | — | — | |
| Proposed-LMNoise rate=0.4x, Epoch selection=Last2023.02 | 59.19 | — | — | — | — | — | — | — | |
| CORES^2Noise rate=0.0, Selection=Last2023.02 | 59.14 | — | — | — | — | — | — | — | |
| CORES^2Noise rate=0.0, Epoch selection=Last2023.02 | 59.14 | — | — | — | — | — | — | — | |
| M-DYRNoise rate=0.1, Selection=Last2023.02 | 59.11 | — | — | — | — | — | — | — | |
| DE-CENoise rate=0.4x, Epoch selection=Last2023.02 | 59.05 | — | — | — | — | — | — | — | |
| Proposed-LNoise rate=0.4, Selection=Best2023.02 | 59.02 | — | — | — | — | — | — | — | |
| Proposed-LNoise rate=0.2, Selection=Best2023.02 | 58.96 | — | — | — | — | — | — | — | |
| Proposed-LNoise rate=0.2, Selection=Last2023.02 | 58.87 | — | — | — | — | — | — | — | |
| Proposed-LMNoise rate=0.2y+0.3x, Epoch selection=Best2023.02 | 58.75 | — | — | — | — | — | — | — | |
| Proposed-LMNoise rate=0.2, Selection=Best2023.02 | 58.65 | — | — | — | — | — | — | — | |
| Proposed-LNoise rate=0.4x, Epoch selection=Last2023.02 | 58.62 | — | — | — | — | — | — | — | |
| M-DYRNoise rate=0.3x, Epoch selection=Last2023.02 | 58.44 | — | — | — | — | — | — | — | |
| M-DYRNoise rate=0.2, Selection=Last2023.02 | 58.25 | — | — | — | — | — | — | — | |
| DYRNoise rate=0.1, Selection=Last2023.02 | 58.13 | — | — | — | — | — | — | — | |
| DE-CENoise rate=0.1, Selection=Last2023.02 | 58.06 | — | — | — | — | — | — | — | |
| Proposed-LMNoise rate=0.2, Selection=Last2023.02 | 58.01 | — | — | — | — | — | — | — | |
| DYRNoise rate=0.4x, Epoch selection=Last2023.02 | 58.01 | — | — | — | — | — | — | — | |
| Proposed-LMNoise rate=0.2y+0.3x, Epoch selection=Last2023.02 | 57.82 | — | — | — | — | — | — | — | |
| Proposed-LNoise rate=0.2y+0.3x, Epoch selection=Best2023.02 | 57.62 | — | — | — | — | — | — | — | |
| CORES^2Noise rate=0.3x, Epoch selection=Best2023.02 | 57.22 | — | — | — | — | — | — | — | |
| M-DYRNoise rate=0.4x, Epoch selection=Last2023.02 | 57.21 | — | — | — | — | — | — | — | |
| CORES^2Noise rate=0.1, Selection=Best2023.02 | 57.15 | — | — | — | — | — | — | — | |
| CORES^2Noise rate=0.1, Selection=Last2023.02 | 57 | — | — | — | — | — | — | — | |
| CORES^2Noise rate=0.3x, Epoch selection=Last2023.02 | 56.35 | — | — | — | — | — | — | — | |
| DYRNoise rate=0.2, Selection=Best2023.02 | 56.33 | — | — | — | — | — | — | — | |
| Proposed-LNoise rate=0.2y+0.3x, Epoch selection=Last2023.02 | 56.31 | — | — | — | — | — | — | — | |
| Proposed-LMNoise rate=0.4, Selection=Best2023.02 | 56.21 | — | — | — | — | — | — | — | |
| Proposed-LNoise rate=0.4, Selection=Last2023.02 | 55.98 | — | — | — | — | — | — | — | |
| Proposed-LMNoise rate=0.4, Selection=Last2023.02 | 55.84 | — | — | — | — | — | — | — | |
| DE-CENoise rate=0.2, Selection=Best2023.02 | 55.82 | — | — | — | — | — | — | — | |
| CORES^2Noise rate=0.4x, Epoch selection=Best2023.02 | 55.67 | — | — | — | — | — | — | — |