Adversarial Attack with Pattern Replacement
arXiv:1911.10875v1
Originality Synthesis-oriented
AI Analysis
This work addresses adversarial vulnerability in neural networks, but it appears incremental as it applies a known generative approach to a specific dataset.
The authors tackled the problem of adversarial attacks by proposing a generative model that replaces input patterns with subtly generated ones from other classes, demonstrating its effectiveness by attacking a CNN on MNIST.
We propose a generative model for adversarial attack. The model generates subtle but predictive patterns from the input. To perform an attack, it replaces the patterns of the input with those generated based on examples from some other class. We demonstrate our model by attacking CNN on MNIST.