CycleGAN, a Master of Steganography

2017/12/08 by Casey Chu, Andrey Zhmoginov, Chu, Casey +4 · 2 voices · 26 citations
Computer Science · Mathematics · #Advanced Steganography and Watermarking Techniques #Computer Vision and Pattern Recognition (cs.CV) #Digital Media Forensic Detection #FOS: Computer and information sciences #Generative Adversarial Networks and Image Synthesis #Machine Learning (cs.LG) #Machine Learning (stat.ML) #cs.CV #cs.LG #stat.ML

paper · pdf · doi:10.48550/arxiv.1712.02950

NIPS 2017, workshop on Machine Deception

openalex publication_date 2017/12/08 · arxiv published 2017/12/08 · arxiv created 2017/12/16 · arxiv updated 2017/12/19 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

CycleGAN (Zhu et al. 2017) is one recent successful approach to learn a transformation between two image distributions. In a series of experiments, we demonstrate an intriguing property of the model: CycleGAN learns to "hide" information about a source image into the images it generates in a nearly imperceptible, high-frequency signal. This trick ensures that the generator can recover the original sample and thus satisfy the cyclic consistency requirement, while the generated image remains realistic. We connect this phenomenon with adversarial attacks by viewing CycleGAN's training procedure as training a generator of adversarial examples and demonstrate that the cyclic consistency loss causes CycleGAN to be especially vulnerable to adversarial attacks.

Citations

• A study of the effect of JPG compression on adversarial images

Cited by

• Robust and Generalizable Background Subtraction on Images of Calorimeter Jets using Unsupervised Generative Learning
• BrainPuzzle: Hybrid Physics and Data-Driven Reconstruction for Transcranial Ultrasound Tomography
• VaseVQA-3D: Benchmarking 3D VLMs on Ancient Greek Pottery
• A Chain-of-thought Reasoning Breast Ultrasound Dataset Covering All Histopathology Categories
• CYCLE-INSTRUCT: Fully Seed-Free Instruction Tuning via Dual Self-Training and Cycle Consistency
• Independent Encoder for Deep Hierarchical Unsupervised Image-to-Image Translation
• MCMI: Multi-Cycle Image Translation with Mutual Information Constraints
• Augmented CycleGAN: Learning Many-to-Many Mappings from Unpaired Data
• Latent Translation: Crossing Modalities by Bridging Generative Models
• Generative Models for Security: Attacks, Defenses, and Opportunities
• Variational Mixture-of-Experts Autoencoders for Multi-Modal Deep Generative Models
• 2D Instance Editing in 3D Space
• Generative Reversible Data Hiding by Image to Image Translation via GANs
• AD-GAN: End-to-end Unsupervised Nuclei Segmentation with Aligned Disentangling Training
• Generating Object Stamps
• Irregular Convolutional Auto-Encoder on Point Clouds
• Out of Control -- Why Alignment Needs Formal Control Theory (and an Alignment Control Stack)
• Semantically Robust Unpaired Image Translation for Data with Unmatched Semantics Statistics
• Built-in Vulnerabilities to Imperceptible Adversarial Perturbations
• RL-CycleGAN: Reinforcement Learning Aware Simulation-To-Real
• On Memory Mechanism in Multi-Agent Reinforcement Learning
• IMPLICITSTAINER: Resolution Agnostic Data-Efficient Virtual Staining Using Neural Implicit Functions
• Self-Supervised Equivariant Scene Synthesis from Video
• EncryptGAN: Image Steganography with Domain Transform
• Generating and Aligning from Data Geometries with Generative Adversarial\n Networks
• A Knowledge-guided Adversarial Defense for Resisting Malicious Visual Manipulation

Discussions

• CycleGAN, a Master of Steganography [2017] [hn, 3 points, 0 comments]
• CycleGAN, a Master of Steganography [hn, 3 points, 0 comments]

Related

• Steganography GAN: Cracking Steganography with Cycle Generative Adversarial Networks
• Cycle-Consistent Adversarial GAN: the integration of adversarial attack and defense
• MMDStegNet: An Adversarial Steganography Framework With Maximum Mean Discrepancy Regularization
• Universal Adversarial Perturbations Through the Lens of Deep Steganography: Towards A Fourier Perspective
• Synchronized Detection and Recovery of Steganographic Messages with Adversarial Learning