Research map / Learning theory and algorithms / Prediction and decision losses
Perfectly Secure Steganography Using Minimum Entropy Coupling
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use iMEC for encoding instead of arithmetic coding, Meteor, or ADG | Information-theoretic steganography · GPT-2, WaveRNN, and Image Transformer covertext distributions | encoding efficiency | Helped |
| Use iMEC instead of arithmetic coding or ADG | Information-theoretic steganography · Covertext distributions evaluated | KL divergence between stegotext and covertext | Helped |
| Use arithmetic coding or ADG instead of iMEC | Information-theoretic steganography · Covertext distributions evaluated | KL divergence between stegotext and covertext | Hurt |
| Use iMEC with 20-bit blocks rather than smaller iMEC blocks (ablation) | iMEC · GPT-2, WaveRNN, and Image Transformer covertext distributions | encoding efficiency | Helped |
| Increase iMEC block size from 10 bits to 16 or 20 bits (ablation) | iMEC · Covertext distributions evaluated | encoding time: 10 bit blocks require an order of magnitude less time than model inference; 16 bit blocks require the same order of magn | Hurt |
| Use iMEC with a 16-bit block size (ablation) | iMEC · Covertext distributions evaluated | encoding efficiency | Mixed |
| Use iMEC with a 10-bit block size (ablation) | iMEC · Covertext distributions evaluated | encoding efficiency | Mixed |
| Use arithmetic coding instead of iMEC with 10- or 16-bit blocks | Information-theoretic steganography · Uniform covertext | encoding efficiency | Helped |
| Use ADG instead of other methods | Information-theoretic steganography · Uniform covertext | encoding efficiency | Helped |
| Use ADG instead of other methods | Information-theoretic steganography · Non-uniform covertext distributions, especially WaveRNN | encoding efficiency | Hurt |
| Use Meteor instead of iMEC variants or arithmetic coding | Information-theoretic steganography · Covertext distributions evaluated | encoding efficiency | Hurt |
| Use Meteor:reorder instead of iMEC with 10-bit blocks | Information-theoretic steganography · GPT-2 90% | encoding efficiency | Helped |
Closest papers by meaning
- SSHR: More Secure Generative Steganography with High-Quality Revealed Secret Images (ICML 2025 · similarity 0.87)
- StegaStyleGAN: Towards Generic and Practical Generative Image Steganography (AAAI 2024 · similarity 0.87)
- Provably Secure Image Robust Steganography via Cross-modal Error Correction (AAAI 2025 · similarity 0.86)
- StegFormer: Rebuilding the Glory of Autoencoder-Based Steganography (AAAI 2024 · similarity 0.85)
- MDDM: Practical Message-Driven Generative Image Steganography Based on Diffusion Models (ICML 2025 · similarity 0.84)
- StegoGAN: Leveraging Steganography for Non-Bijective Image-to-Image Translation (CVPR 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.