Image Classification on Waterbird modified (test)
15.13Balanced ErrorAutoBalance
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| AutoBalancevalidation loss lambda=0.1, training objective=optimized L_fair2022.01 | 15.13 | 30.33 | 4.25 | |
| DROLoss function=DRO2022.01 | 16.47 | 32.67 | 6.91 | |
| CE_deoLoss function=CE_deo2022.01 | 19.29 | 35.17 | 25.25 | |
| CE + CE_deo blendlambda=0.1, Loss function=0.1 * CE + 0.9 * CE_deo2022.01 | 20.06 | 31.67 | 26.25 | |
| CE_balLoss function=CE_bal2022.01 | 20.83 | 36.67 | 20.25 | |
| Group-LA lossLoss function=Group-LA loss2022.01 | 22.83 | 40.5 | 29.33 | |
| Cross entropy (CE)Loss function=Cross entropy (CE)2022.01 | 23.38 | 43.25 | 33.75 |