← Back to Basic Science

GRU Encoder-Decoder

2014
Computer Science (theoretical)Machine Learning TheoryArchitecture Componentfoundational

RNN encoder-decoder using gated recurrent units to map a variable-length input sequence to a variable-length output sequence for statistical machine translation (Cho et al., "Learning Phrase Representations using RNN Encoder-Decoder for Statistical Machine Translation," arXiv:1406.1078, 2014). A simpler gated-recurrence alternative to LSTM, and the architecture Bahdanau et al.'s attention mechanism was built on top of.

Originators

  • Cho, K.
  • van Merrienboer, B.
  • Gulcehre, C.
  • Bahdanau, D.
  • Bougares, F.
  • Schwenk, H.
  • Bengio, Y.

Landmark Paper

W2157331557 ↗
Not retracted (OpenAlex)

Checked 2026-09-19 — interim signal only, see docs/BASIC_ROADMAP.md Phase 10

Connections

  • is component of Attention Mechanism
    basis: reasoned

    Bahdanau, Cho & Bengio's attention mechanism (ref [2]) was introduced inside the RNN encoder-decoder architecture from Cho et al. 2014 (ref [5]); the encoder-decoder is the structural context attention was added to, not a separate downstream consumer of it.