AAAI Publications, Workshops at the Thirty-Second AAAI Conference on Artificial Intelligence

Font Size: 
Denoising Dictionary Learning Against Adversarial Perturbations
John Mitro, Derek Bridge, Steven Prestwich

Last modified: 2018-06-20

Abstract


We propose denoising dictionary learning (DDL), a simple yet effective technique as a protection measure against adversarial perturbations. We examined demising dictionary learning on MNIST andCIFAR10 perturbed under two different perturbation techniques, fast gradient sign (FGSM) and jacobian saliency maps (JSMA). We evaluated it against five different deep neural networks (DNN) representing the building blocks of most recent architectures indicating a successive progression of model complexity of each other. We show that each model tends to capture different representations based on their architecture. For each model we recorded its accuracy both on the perturbed test data previously misclassified with high confidence and on the denoised one after the reconstruction using dictionary learning. The reconstruction quality of each data point is assessed by means of PSNR (Peak Signal to Noise Ratio) and Structure Similarity Index (SSI). We show that after applying (DDL) the reconstruction of the original data point from a noisy sample results in a correct prediction with high confidence.

Keywords


adversarial examples; deep neural networks; robustness; denoise; dictionary learning; security

Full Text: PDF