Class-Incremental Learning on ImageNet-100 10 tasks
85.7AccuracyHydraCIL
Evaluation results
| Method | Links | |
|---|---|---|
| HydraCILk (number of prototypes)=20, Time (min)=7.11, Energy (kWh)=0.021, Emissions (kg CO2-eq)=0.0042026.06 | 85.7 | |
| HydraCILk (number of prototypes)=2, Time (min)=6.72, Energy (kWh)=0.019, Emissions (kg CO2-eq)=0.0032026.06 | 85.66 | |
| HydraCILk (number of prototypes)=5, Time (min)=6.68, Energy (kWh)=0.020, Emissions (kg CO2-eq)=0.0032026.06 | 85.5 | |
| CIFNetTime (min)=70.02, Energy (kWh)=0.264, Emissions (kg CO2-eq)=0.0462026.06 | 77.92 | |
| RMM-FOSTERTime (min)=757.55, Energy (kWh)=5.186, Emissions (kg CO2-eq)=0.9032026.06 | 66.1 | |
| DERTime (min)=742.22, Energy (kWh)=5.328, Emissions (kg CO2-eq)=0.9272026.06 | 64.4 | |
| Fine-tuneTime (min)=184.20, Energy (kWh)=1.152, Emissions (kg CO2-eq)=0.2012026.06 | 9.34 |