BBN+
motivation
观测用BBN的模式训练出来的模型, 配上不同的\(\alpha\), 结果会如何.
settings
| Attribute | Value |
|---|---|
| attack | pgd-linf |
| batch_size | 128 |
| beta1 | 0.9 |
| beta2 | 0.999 |
| dataset | cifar10 |
| description | AT=True-0.0=default-sgd-0.1=pgd-linf-0.0314-0.25-10=128=default |
| epochs | 100 |
| epsilon | 0.03137254901960784 |
| eva_alpha | 0.0 |
| learning_policy | [50, 75] x 0.1 |
| loss | cross_entropy |
| lr | 0.1 |
| model | resnet32 |
| momentum | 0.9 |
| norm_cls | True |
| optimizer | sgd |
| progress | False |
| resume | False |
| seed | 1 |
| stats_log | False |
| steps | 10 |
| stepsize | 0.25 |
| transform | default |
| weight_decay | 0.0005 |
results
| Accuracy | Robustness | |
|---|---|---|
| parabolic decay |  |
 |
| fixed_alpha=0.5 |  |
 |
| fixed_alpha=0.5, 同时对抗样本也是采用alpha=0.5生成的 |  |
 |
太奇怪了, 为啥会发生这种事情?
