Bitstream Collisions in Neural Image Compression via Adversarial Perturbations

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Madden, Jordan, Dorje, Lhamo, Li, Xiaohua
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866910893155024896
author Madden, Jordan
Dorje, Lhamo
Li, Xiaohua
author_facet Madden, Jordan
Dorje, Lhamo
Li, Xiaohua
contents Neural image compression (NIC) has emerged as a promising alternative to classical compression techniques, offering improved compression ratios. Despite its progress towards standardization and practical deployment, there has been minimal exploration into it's robustness and security. This study reveals an unexpected vulnerability in NIC - bitstream collisions - where semantically different images produce identical compressed bitstreams. Utilizing a novel whitebox adversarial attack algorithm, this paper demonstrates that adding carefully crafted perturbations to semantically different images can cause their compressed bitstreams to collide exactly. The collision vulnerability poses a threat to the practical usability of NIC, particularly in security-critical applications. The cause of the collision is analyzed, and a simple yet effective mitigation method is presented.
format Preprint
id arxiv_https___arxiv_org_abs_2503_19817
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Bitstream Collisions in Neural Image Compression via Adversarial Perturbations
Madden, Jordan
Dorje, Lhamo
Li, Xiaohua
Cryptography and Security
Artificial Intelligence
Neural image compression (NIC) has emerged as a promising alternative to classical compression techniques, offering improved compression ratios. Despite its progress towards standardization and practical deployment, there has been minimal exploration into it's robustness and security. This study reveals an unexpected vulnerability in NIC - bitstream collisions - where semantically different images produce identical compressed bitstreams. Utilizing a novel whitebox adversarial attack algorithm, this paper demonstrates that adding carefully crafted perturbations to semantically different images can cause their compressed bitstreams to collide exactly. The collision vulnerability poses a threat to the practical usability of NIC, particularly in security-critical applications. The cause of the collision is analyzed, and a simple yet effective mitigation method is presented.
title Bitstream Collisions in Neural Image Compression via Adversarial Perturbations
topic Cryptography and Security
Artificial Intelligence
url https://arxiv.org/abs/2503.19817