Multi-Task Learning on Taskonomy Tiny
3.7ΔT1Multi-Task
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Multi-Task# Params=-80.02019.11 | 3.7 | 1.6 | 4.5 | 0 | 4.2 | 1.1 | |
| AdaShare# Params=-80.02019.11 | 2.3 | 0.7 | 4.5 | 3 | 5.7 | 1.1 | |
| Cross-Stitch# Params=0.02019.11 | 0.9 | 4 | 0 | 1 | 2.4 | 1.3 | |
| Sluice# Params=0.02019.11 | -3.7 | -1.7 | -9.1 | 0.5 | 2.4 | 2.3 | |
| NDDR-CNN# Params=+8.22019.11 | -4.2 | -1 | 4.5 | 2 | 4.2 | 0.7 | |
| MTAN# Params=-9.82019.11 | -8 | 2.8 | 4.5 | 0 | 2.8 | 2.5 | |
| DEN# Params=-77.62019.11 | -28.2 | 3 | -22.7 | 2.5 | 4.2 | 9.4 |