Detecting and Corrupting Convolution-based Unlearnable Examples
This work addresses a critical security issue in machine learning by providing the first effective defense against a new type of unlearnable examples, which is incremental but essential for robust model training.
The paper tackles the problem of convolution-based unlearnable examples (UEs) that degrade model performance by introducing class-wise multiplicative convolutional noise, and proposes a detection method (EPD) and a defense scheme (COIN) that outperforms 11 state-of-the-art defenses on CIFAR and ImageNet datasets.
Convolution-based unlearnable examples (UEs) employ class-wise multiplicative convolutional noise to training samples, severely compromising model performance. This fire-new type of UEs have successfully countered all defense mechanisms against UEs. The failure of such defenses can be attributed to the absence of norm constraints on convolutional noise, leading to severe blurring of image features. To address this, we first design an Edge Pixel-based Detector (EPD) to identify convolution-based UEs. Upon detection of them, we propose the first defense scheme against convolution-based UEs, COrrupting these samples via random matrix multiplication by employing bilinear INterpolation (COIN) such that disrupting the distribution of class-wise multiplicative noise. To evaluate the generalization of our proposed COIN, we newly design two convolution-based UEs called VUDA and HUDA to expand the scope of convolution-based UEs. Extensive experiments demonstrate the effectiveness of detection scheme EPD and that our defense COIN outperforms 11 state-of-the-art (SOTA) defenses, achieving a significant improvement on the CIFAR and ImageNet datasets.