Reinforcement Learning on MountainCar (Wall-clock training time)
1.4Training Time (min)Independent QHD
Evaluation results
| Method | Links | |
|---|---|---|
| Independent QHDClient heterogeneity setup=Heterogeneous, Number of episodes=600, Number of clients=52026.05 | 1.4 | |
| FedQHDClient heterogeneity setup=Heterogeneous, Number of episodes=600, Number of clients=52026.05 | 1.4 | |
| Truncate FedAvg-QHDClient heterogeneity setup=Heterogeneous, Number of episodes=600, Number of clients=52026.05 | 1.7 | |
| Truncate FedAvg-QHDClient heterogeneity setup=Homogeneous, Number of episodes=600, Number of clients=52026.05 | 7.6 | |
| Oracle QHD†Client heterogeneity setup=Homogeneous, Number of episodes=600, Number of clients=52026.05 | 7.7 | |
| Oracle QHD†Client heterogeneity setup=Heterogeneous, Number of episodes=600, Number of clients=52026.05 | 8.5 | |
| Independent QHDClient heterogeneity setup=Homogeneous, Number of episodes=600, Number of clients=52026.05 | 12.7 | |
| FedQHDClient heterogeneity setup=Homogeneous, Number of episodes=600, Number of clients=52026.05 | 17.2 | |
| Distillation FedDQNClient heterogeneity setup=Homogeneous, Number of episodes=600, Number of clients=52026.05 | 28.1 | |
| Distillation FedDQNClient heterogeneity setup=Heterogeneous, Number of episodes=600, Number of clients=52026.05 | 52.9 | |
| Oracle DQN†Client heterogeneity setup=Homogeneous, Number of episodes=600, Number of clients=52026.05 | 67.8 | |
| FedAvg-DQNClient heterogeneity setup=Homogeneous, Number of episodes=600, Number of clients=52026.05 | 85.6 | |
| Oracle DQN†Client heterogeneity setup=Heterogeneous, Number of episodes=600, Number of clients=52026.05 | 101.5 |