Federated Unlearning on Fashion-MNIST (val)
73.85Pretrain Accuracyf-FUM (KL-KL)
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| f-FUM (KL-KL)N=10, Backbone=MINILENET, Unlearning Scenario=Client-level (80% backdoored data removed)2026.02 | 73.85 | 73.1 | |
| f-FUM (KL-JS)N=10, Backbone=MINILENET, Unlearning Scenario=Client-level (80% backdoored data removed)2026.02 | 73.85 | 75.13 | |
| f-FUM (KL-χ2)N=10, Backbone=MINILENET, Unlearning Scenario=Client-level (80% backdoored data removed)2026.02 | 73.85 | 73.82 | |
| NoTN=10, Backbone=MINILENET, Unlearning Scenario=Client-level (80% backdoored data removed)2026.02 | 73.85 | 46.82 | |
| HalimiN=10, Backbone=MINILENET, Unlearning Scenario=Client-level (80% backdoored data removed)2026.02 | 73.85 | 73.32 | |
| MoDEN=10, Backbone=MINILENET, Unlearning Scenario=Client-level (80% backdoored data removed)2026.02 | 73.85 | 72.07 | |
| f-FUM (KL-KL)N=5, Backbone=MINILENET, Unlearning Scenario=Client-level (80% backdoored data removed)2026.02 | 31.32 | 39.55 | |
| f-FUM (KL-JS)N=5, Backbone=MINILENET, Unlearning Scenario=Client-level (80% backdoored data removed)2026.02 | 31.32 | 55.93 | |
| f-FUM (KL-χ2)N=5, Backbone=MINILENET, Unlearning Scenario=Client-level (80% backdoored data removed)2026.02 | 31.32 | 48.48 | |
| NoTN=5, Backbone=MINILENET, Unlearning Scenario=Client-level (80% backdoored data removed)2026.02 | 31.32 | 37.45 | |
| HalimiN=5, Backbone=MINILENET, Unlearning Scenario=Client-level (80% backdoored data removed)2026.02 | 31.32 | 54.02 | |
| MoDEN=5, Backbone=MINILENET, Unlearning Scenario=Client-level (80% backdoored data removed)2026.02 | 31.32 | 64.3 |