Saved in:
Bibliographic Details
Main Authors: Xu, Changming, Banerjee, Debangshu, Vasisht, Deepak, Singh, Gagandeep
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2504.11831
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912348817588224
author Xu, Changming
Banerjee, Debangshu
Vasisht, Deepak
Singh, Gagandeep
author_facet Xu, Changming
Banerjee, Debangshu
Vasisht, Deepak
Singh, Gagandeep
contents Variational Autoencoders (VAEs) have become increasingly popular and deployed in safety-critical applications. In such applications, we want to give certified probabilistic guarantees on performance under adversarial attacks. We propose a novel method, CIVET, for certified training of VAEs. CIVET depends on the key insight that we can bound worst-case VAE error by bounding the error on carefully chosen support sets at the latent layer. We show this point mathematically and present a novel training algorithm utilizing this insight. We show in an extensive evaluation across different datasets (in both the wireless and vision application areas), architectures, and perturbation magnitudes that our method outperforms SOTA methods achieving good standard performance with strong robustness guarantees.
format Preprint
id arxiv_https___arxiv_org_abs_2504_11831
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Support is All You Need for Certified VAE Training
Xu, Changming
Banerjee, Debangshu
Vasisht, Deepak
Singh, Gagandeep
Machine Learning
Variational Autoencoders (VAEs) have become increasingly popular and deployed in safety-critical applications. In such applications, we want to give certified probabilistic guarantees on performance under adversarial attacks. We propose a novel method, CIVET, for certified training of VAEs. CIVET depends on the key insight that we can bound worst-case VAE error by bounding the error on carefully chosen support sets at the latent layer. We show this point mathematically and present a novel training algorithm utilizing this insight. We show in an extensive evaluation across different datasets (in both the wireless and vision application areas), architectures, and perturbation magnitudes that our method outperforms SOTA methods achieving good standard performance with strong robustness guarantees.
title Support is All You Need for Certified VAE Training
topic Machine Learning
url https://arxiv.org/abs/2504.11831