[WIP] Implementation of Maximally Interfered Retrevial for Continual Learning
-
Implement iid online
-
Implement iid online offline
-
Implement fine-tuning
| Dataset | Acc | Forgetting |
|---|---|---|
| Split-MNIST | 0.19 | 0.994 |
| Permute-MNIST | 0.71 | 0.28 |
| CIFAR-10 |
- Implement ER
| Dataset | Acc | Forgetting |
|---|---|---|
| Split-MNIST | ||
| Permute-MNIST | ||
| CIFAR-10 |
- Implement ER-MIR
Method: SM-1
| Dataset | Acc | Forgetting |
|---|---|---|
| Split-MNIST | ||
| Permute-MNIST | ||
| CIFAR-10 |
Method: SM-2
| Dataset | Acc | Forgetting |
|---|---|---|
| Split-MNIST | ||
| Permute-MNIST | ||
| CIFAR-10 |
- Implement GEN-MIR
| Dataset | Acc | Forgetting |
|---|---|---|
| Split-MNIST | ||
| Permute-MNIST | ||
| CIFAR-10 |