Classification on MNIST (test) (Bound Loss, Test Error, KL Divergence)
0.028Bound Loss (0-1)Perez-Ortiz et al. (2021b)
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Perez-Ortiz et al. (2021b)Setup=1 client, Prior=Learnt, Objective=f12023.10 | 0.028 | 23 | 1 | |
| Perez-Ortiz et al. (2021b)Setup=1 client, Prior=Learnt, Objective=f22023.10 | 0.028 | 20 | 1 | |
| GenFLSetup=100 users, KL Penalty=No penalty, Prior=Learnt, Objective=f12023.10 | 0.039 | 30 | 1 | |
| GenFLSetup=100 users, KL Penalty=No penalty, Prior=Learnt, Objective=f22023.10 | 0.04 | 30 | 1 | |
| Perez-Ortiz et al. (2021b)Setup=1 client, Prior=Random, Objective=f22023.10 | 0.316 | 92 | 138 | |
| Perez-Ortiz et al. (2021b)Setup=1 client, Prior=Random, Objective=f12023.10 | 0.33 | 141 | 81 | |
| GenFLSetup=100 users, KL Penalty=0.1, Prior=Random, Objective=f12023.10 | 0.333 | 123 | 107 | |
| GenFLSetup=100 users, KL Penalty=0.1, Prior=Random, Objective=f22023.10 | 0.342 | 90 | 163 | |
| GenFLSetup=100 users, KL Penalty=0.1, Prior=Learnt, Objective=f12023.10 | 61 | 30 | 1 | |
| GenFLSetup=100 users, KL Penalty=0.1, Prior=Learnt, Objective=f22023.10 | 88 | 29 | 2 | |
| GenFLSetup=100 users, KL Penalty=No penalty, Prior=Random, Objective=f22023.10 | 408 | 251 | 41 | |
| GenFLSetup=100 users, KL Penalty=No penalty, Prior=Random, Objective=f12023.10 | 415 | 256 | 39 |