# Dynamic Past and Future for Neural Machine Translation

arXiv: https://arxiv.org/abs/1904.09646 · HTML: https://mlsanity.com/paper/1904.09646/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 21 Apr 2019.

## Built on by (most cited first, 5 of 5)

- [Selective Knowledge Distillation for Neural Machine Translation](https://mlsanity.com/paper/2105.12967/) (2021)
- [Faithfulness in Natural Language Generation: A Systematic Survey of Analysis, Evaluation and Optimization Methods](https://mlsanity.com/paper/2203.05227/) (2022)
- [Turning Fixed to Adaptive: Integrating Post-Evaluation into Simultaneous Machine Translation](https://mlsanity.com/paper/2210.11900/) (2022)
- [Guiding Teacher Forcing with Seer Forcing for Neural Machine Translation](https://mlsanity.com/paper/2106.06751/) (2021)
- [Modeling Coverage for Non-Autoregressive Neural Machine Translation](https://mlsanity.com/paper/2104.11897/) (2021)

## Abstract (arXiv)

Previous studies have shown that neural machine translation (NMT) models can benefit from explicitly modeling translated (Past) and untranslated (Future) to groups of translated and untranslated contents through parts-to-wholes assignment. The assignment is learned through a novel variant of routing-by-agreement mechanism (Sabour et al., 2017), namely em Guided Dynamic Routing, where the translating status at each decoding step em guides the routing process to assign each source word to its associated group (i.e., translated or untranslated content) represented by a capsule, enabling translation to be made from holistic context. Experiments show that our approach achieves substantial improvements over both RNMT and Transformer by producing more adequate translations. Extensive analysis demonstrates that our method is highly interpretable, which is able to recognize the translated and untranslated contents as expected.