Amicable Aid: Perturbing Images to Improve Classification Performance
This work addresses the problem of enhancing model robustness and accuracy in image classification, offering a counterpoint to adversarial attacks, though it is incremental in applying perturbation concepts beneficially.
The paper introduces 'amicable aid', a novel paradigm where image perturbations are used to improve classification performance by increasing confidence and correcting misclassifications, even with large perturbations that make images unrecognizable to humans.
While adversarial perturbation of images to attack deep image classification models pose serious security concerns in practice, this paper suggests a novel paradigm where the concept of image perturbation can benefit classification performance, which we call amicable aid. We show that by taking the opposite search direction of perturbation, an image can be modified to yield higher classification confidence and even a misclassified image can be made correctly classified. This can be also achieved with a large amount of perturbation by which the image is made unrecognizable by human eyes. The mechanism of the amicable aid is explained in the viewpoint of the underlying natural image manifold. Furthermore, we investigate the universal amicable aid, i.e., a fixed perturbation can be applied to multiple images to improve their classification results. While it is challenging to find such perturbations, we show that making the decision boundary as perpendicular to the image manifold as possible via training with modified data is effective to obtain a model for which universal amicable perturbations are more easily found.