Why don't people use character-level machine translation?

Libovický, Jindřich; Schmid, Helmut; Fraser, Alexander

doi:10.48550/arxiv.2110.08191

Cited by 4 publications

(7 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Indeed, the curves for both BLEU (1a) and gender coverage (1b) have a rapid and steady initial increase, 11 which starts to level off around the 20th ckp. 12 Also, the BLEU trends reveal a divide across models (BPE>CHAR) that remains visible over the 9 Contemporary to our submission, Libovickỳ et al (2021) show that en-de MT systems based on character-level segmentation have an edge -with respect to BPE -in terms of gender accuracy on the WinoMT benchmark . Their results, however, do not distinguish between feminine and masculine translation capabilities.…”

Section: Overall Resultsmentioning

confidence: 68%

Proceedings of the 4th Workshop on Gender Bias in Natural Language Processing (GeBNLP)

2022

View full text Add to dashboard Cite

Warning: This work contains strong and offensive language, sometimes uncensored.To tackle the rising phenomenon of hate speech, efforts have been made towards data curation and analysis. When it comes to analysis of bias, previous work has focused predominantly on race. In our work, we further investigate bias in hate speech datasets along racial, gender and intersectional axes. We identify strong bias against African American English (AAE), masculine and AAE+Masculine tweets, which are annotated as disproportionately more hateful and offensive than from other demographics. We provide evidence that BERT-based models propagate this bias and show that balancing the training data for these protected attributes can lead to fairer models with regards to gender, but not race.

show abstract

Section: Overall Resultsmentioning

confidence: 68%

Proceedings of the 4th Workshop on Gender Bias in Natural Language Processing (GeBNLP)

2022

View full text Add to dashboard Cite

show abstract

“…Contemporary to our submission,Libovickỳ et al (2021) show that en-de MT systems based on character-level segmentation have an edge -with respect to BPE -in terms of gender accuracy on the WinoMT benchmark(Stanovsky et al, 2019). Their results, however, do not distinguish between feminine and masculine translation capabilities.10 For the sake of our analysis across epochs, we do not generate our final systems by averaging the 5 models around the best ckp as in andSavoldi et al (2022).…”

mentioning

confidence: 86%

On the Dynamics of Gender Learning in Speech Translation

Savoldi¹,

Gaido²,

Bentivogli³

et al. 2022

Proceedings of the 4th Workshop on Gender Bias in Natural Language Processing (GeBNLP)

View full text Add to dashboard Cite

Due to the complexity of bias and the opaque nature of current neural approaches, there is a rising interest in auditing language technologies. In this work, we contribute to such a line of inquiry by exploring the emergence of gender bias in Speech Translation (ST). As a new perspective, rather than focusing on the final systems only, we examine their evolution over the course of training. In this way, we are able to account for different variables related to the learning dynamics of gender translation, and investigate when and how gender divides emerge in ST. Accordingly, for three language pairs (en -> es, fr, it) we compare how ST systems behave for masculine and feminine translation at several levels of granularity. We find that masculine and feminine curves are dissimilar, with the feminine one being characterized by a more erratic behaviour and late improvements over the course of training. Also, depending on the considered phenomena, their learning trends can be either antiphase or parallel. Overall, we show how such a progressive analysis can inform on the reliability and time-wise acquisition of gender, which is concealed by static evaluations and standard metrics.

show abstract

“…We now explain the architecture used in our experiments. We build off of the previous work by using the CNN downsampling architecture followed by the Transformer and using Libovickỳ et al (2021)'s two-step decoding with an LSTM for upsampling. This previous work was only applied to fixed-length downsampling and upsampling, however the aforementioned WDD and SDD methods require variable-length downsampling and upsampling.…”

Section: Architecturementioning

confidence: 99%

“…To alleviate the problem of training time, several methods have been proposed to initially downsample characters into shorter sequences, which are then fed into the encoder or decoder. For discriminative tasks, these can be applied without any loss in performance (Tay et al, 2021), however for generative tasks like NMT, the performance is either untested or lacking when compared to character models without downsampling (Libovickỳ et al, 2021).…”

Section: Introductionmentioning

confidence: 99%

Subword-Delimited Downsampling for Better Character-Level Translation

Edman¹,

Toral²,

Noord³

2022

Findings of the Association for Computational Linguistics: EMNLP 2022

View full text Add to dashboard Cite

Subword-level models have been the dominant paradigm in NLP. However, character-level models have the benefit of seeing each character individually, providing the model with more detailed information that ultimately could lead to better models. Recent works have shown character-level models to be competitive with subword models, but costly in terms of time and computation. Character-level models with a downsampling component alleviate this, but at the cost of quality, particularly for machine translation. This work analyzes the problems of previous downsampling methods and introduces a novel downsampling method which is informed by subwords. This new downsampling method not only outperforms existing downsampling methods, showing that downsampling characters can be done without sacrificing quality, but also leads to promising performance compared to subword models for translation.

show abstract

Why don't people use character-level machine translation?

Cited by 4 publications

References 42 publications

Proceedings of the 4th Workshop on Gender Bias in Natural Language Processing (GeBNLP)

Proceedings of the 4th Workshop on Gender Bias in Natural Language Processing (GeBNLP)

On the Dynamics of Gender Learning in Speech Translation

Subword-Delimited Downsampling for Better Character-Level Translation

Contact Info

Product

Resources

About