Memory Defense: More Robust Classification via a Memory-Masking Autoencoder

Kavli Affiliate: Dan Luo

| First 5 Authors: Eashan Adhikarla, Dan Luo, Brian D. Davison, ,

| Summary:

Many deep neural networks are susceptible to minute perturbations of images
that have been carefully crafted to cause misclassification. Ideally, a robust
classifier would be immune to small variations in input images, and a number of
defensive approaches have been created as a result. One method would be to
discern a latent representation which could ignore small changes to the input.
However, typical autoencoders easily mingle inter-class latent representations
when there are strong similarities between classes, making it harder for a
decoder to accurately project the image back to the original high-dimensional
space. We propose a novel framework, Memory Defense, an augmented classifier
with a memory-masking autoencoder to counter this challenge. By masking other
classes, the autoencoder learns class-specific independent latent
representations. We test the model’s robustness against four widely used
attacks. Experiments on the Fashion-MNIST & CIFAR-10 datasets demonstrate the
superiority of our model. We make available our source code at GitHub
repository: https://github.com/eashanadhikarla/MemDefense

| Search Query: ArXiv Query: search_query=au:”Dan Luo”&id_list=&start=0&max_results=10

Read More

Leave a Reply