AI Chat AI Image Generator AI Video Text to Speech

Diagnosing Vulnerability of Variational Auto-Encoders to Adversarial Attacks

03/10/2021

∙

by Anna Kuzina, et al.

∙

∙

In this work, we explore adversarial attacks on the Variational Autoencoders (VAE). We show how to modify data point to obtain a prescribed latent code (supervised attack) or just get a drastically different code (unsupervised attack). We examine the influence of model modifications (β-VAE, NVAE) on the robustness of VAEs and suggest metrics to quantify it.

Anna Kuzina
10 publications
Max Welling
167 publications
Jakub M. Tomczak
43 publications

page 1

page 3

research

∙ 03/18/2022

Defending Variational Autoencoders from Adversarial Attacks with MCMC

Variational autoencoders (VAEs) are deep generative models used in vario...

0 Anna Kuzina, et al. ∙

research

∙ 06/12/2018

Adversarial Attacks on Variational Autoencoders

Adversarial attacks are malicious inputs that derail machine-learning mo...

2 George Gondim-Ribeiro, et al. ∙

research

∙ 07/14/2020

Towards a Theoretical Understanding of the Robustness of Variational Autoencoders

We make inroads into understanding the robustness of Variational Autoenc...

16 Alexander Camuto, et al. ∙

research

∙ 05/29/2021

Understanding Instance-based Interpretability of Variational Auto-Encoders

Instance-based interpretation methods have been widely studied for super...

46 Zhifeng Kong, et al. ∙

research

∙ 08/22/2023

Adversarial Attacks on Code Models with Discriminative Graph Patterns

Pre-trained language models of code are now widely used in various softw...

0 Thanh-Dat Nguyen, et al. ∙

research

∙ 03/04/2020

Double Backpropagation for Training Autoencoders against Adversarial Attack

Deep learning, as widely known, is vulnerable to adversarial samples. Th...

0 Chengjin Sun, et al. ∙

research

∙ 12/07/2018

Adversarial Defense of Image Classification Using a Variational Auto-Encoder

Deep neural networks are known to be vulnerable to adversarial attacks. ...

0 Yi Luo, et al. ∙