{
  "id": "rnn-stega",
  "title": "RNN-Stega: Linguistic Steganography Based on Recurrent Neural Networks",
  "authors": [
    "Zhong-Liang Yang",
    "Xiao-Qing Guo",
    "Zi-Ming Chen",
    "Yong-Feng Huang",
    "Yu-Jin Zhang"
  ],
  "author_parts": [
    {
      "given": "Zhong-Liang",
      "family": "Yang"
    },
    {
      "given": "Xiao-Qing",
      "family": "Guo"
    },
    {
      "given": "Zi-Ming",
      "family": "Chen"
    },
    {
      "given": "Yong-Feng",
      "family": "Huang"
    },
    {
      "given": "Yu-Jin",
      "family": "Zhang"
    }
  ],
  "year": 2019,
  "venue_date": "2019-05",
  "venue": "IEEE Transactions on Information Forensics and Security",
  "venue_short": "IEEE TIFS 2019 (online 2018)",
  "publication_type": "journal",
  "publisher": "IEEE",
  "volume": "14",
  "issue": "5",
  "pages": "1280–1295",
  "doi": "10.1109/TIFS.2018.2871746",
  "arxiv_id": null,
  "project_url": "https://github.com/YangzlTHU/Linguistic-Steganography-and-Steganalysis/tree/master/Steganography/RNN-Stega",
  "relation": "foundation",
  "categories": [
    "Earlier neural methods",
    "Huffman coding"
  ],
  "summary": "Generates cover text with a recurrent language model and dynamically encodes candidate words using fixed-length or Huffman variable-length codes. It evaluates generation efficiency, embedding capacity, text quality, and resistance to the tested steganalysis methods.",
  "relationship": "The 2019 Neural Linguistic Steganography paper compares against this work's Huffman approach using the same GPT-2 language model as its own arithmetic method.",
  "limitation": "The paper describes a capacity–quality trade-off and the cost of rebuilding a Huffman tree at each generation step. These empirical results should not be read as a distribution-preservation guarantee.",
  "sources": [
    {
      "url": "https://ieeexplore.ieee.org/document/8470163",
      "evidence": "Publisher record: first online publication on September 24, 2018; May 2019 journal issue, pages, and DOI."
    },
    {
      "url": "https://api.crossref.org/works/10.1109/TIFS.2018.2871746",
      "evidence": "Publisher-deposited complete author spellings, title, May 2019 issue metadata, DOI."
    },
    {
      "url": "https://www.researchgate.net/publication/327852903_RNN-Stega_Linguistic_Steganography_Based_on_Recurrent_Neural_Networks",
      "evidence": "Full manuscript explicitly uploaded by author Zhongliang Yang; section III-B gives FLC/VLC and Huffman rebuilding trade-off; section IV gives evaluations."
    },
    {
      "url": "https://aclanthology.org/D19-1115.pdf",
      "evidence": "Sections 2 and 4 describe the Huffman comparison and shared-model reimplementation."
    },
    {
      "url": "https://raw.githubusercontent.com/YangzlTHU/Linguistic-Steganography-and-Steganalysis/master/Steganography/RNN-Stega/README.md",
      "evidence": "Author-lab description of the RNN-Stega project."
    }
  ],
  "earliest_date": "2018-09-24",
  "earliest_date_label": "First online publication (IEEE)",
  "url": "https://ieeexplore.ieee.org/document/8470163",
  "pdf": null,
  "code": null,
  "category": "foundations",
  "contribution": "Generates cover text with a recurrent language model and dynamically encodes candidate words using fixed-length or Huffman variable-length codes. It evaluates generation efficiency, embedding capacity, text quality, and resistance to the tested steganalysis methods.",
  "canonical_url": "https://steganography.live/reading-list/papers/rnn-stega/"
}