A Shared Task on Bandit Learning for Machine Translation
By Artem Sokolov, Julia Kreutzer, Kellen Sunderland, Pavel Danchenko, Witold Szymaniak, Hagen Fürstenau, Stefan Riezler. 2017
16 citations recorded by the source corpus.
We introduce and describe the results of a novel shared task on bandit learning for machine translation. The task was organized jointly by Amazon and Heidelberg University for the first time at the Second Conference on Machine Translation (WMT 2017). The goal of the task is to encourage research on learning machine translation from weak user feedback instead of human references or post-edits. On each of a sequence of rounds, a machine translation system is required to propose a translation for an input, and receives a real-valued estimate of the quality of the proposed translation for learning. This paper describes the shared task's learning and evaluation setup, using services hosted on Amazon Web Services (AWS), the data and evaluation metrics, and the results of various machine translation architectures and learning protocols.
Subfields: Natural Language Processing
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Natural Language Processing
- Julia Kreutzer
- Stefan Riezler
- Shahram Khadivi
- Evgeny Matusov
- Samuel Kiegeland
- Marta Ruiz Costa-jussà
- Magdalena Biesialska
- Katarzyna Biesialska
Citations in the corpus
- Can Neural Machine Translation be Improved with User Feedback? (2018, background): “Expected Loss (EL) maximizes the expectation of a reward over all source and target sequences: (Kreutzer et al., 2017; Sokolov et al., 2017) R(θ) =Ep(x)pθ(ỹ|x) [∆(ỹ)]”
- Continual Lifelong Learning in Natural Language Processing: A Survey (2020, background): “In a similar vein, there were approaches (Sokolov et al., 2017) to incorporate bandit learners, which implicitly involve domain adaptation and on-line learning, for MT systems.”
- Revisiting the Weaknesses of Reinforcement Learning for Neural Machine Translation (2021, background): “making the distribution peakier) may be used to make policies more deterministic towards the end of training (Sutton and Barto, 1998; Rose, 1998; Sokolov et al., 2017), while we aim to reduce peakiness by increasing the temperature.”