Multi-class Classification on Aloi
0.0458LossRL with rejection-based reward
Evaluation Results
| Method | Links | |
|---|---|---|
| RL with rejection-based rewardArchitecture=176-144-144-80, Number of Parameters=161,6722022.04 | 0.0458 | |
| RL with rejection-based rewardArchitecture=192-112-176-80, Number of Parameters=161,4322022.04 | 0.0461 | |
| MNasNet (reward type 1)Architecture=160-128-112-96, Number of Parameters=163,5442022.04 | 0.0469 | |
| reference architectureArchitecture=160-64-512-64, Number of Parameters=162,0562022.04 | 0.047 | |
| RL with Abs RewardArchitecture=144-112-144-96, Number of Parameters=162,0082022.04 | 0.0473 | |
| MNasNet (reward type 1)Architecture=128-176-128-80, Number of Parameters=153,1922022.04 | 0.0473 | |
| MNasNet (reward type 2)Architecture=112-144-240-64, Number of Parameters=145,9442022.04 | 0.048 | |
| RL with Abs RewardArchitecture=128-96-96-112, Number of Parameters=162,0722022.04 | 0.0488 | |
| MNasNet (reward type 1)Architecture=128-96-96-112, Number of Parameters=162,0722022.04 | 0.0488 | |
| MNasNet (reward type 2)Architecture=160-112-128-64, Number of Parameters=126,3922022.04 | 0.0497 | |
| RL with Abs RewardArchitecture=112-112-96-112, Number of Parameters=161,8162022.04 | 0.0502 | |
| MNasNet (reward type 2)Architecture=256-64-224-32, Number of Parameters=104,2322022.04 | 0.0548 |