Most natural language processing (NLP) learning algorithms require labeled data. While this is given for a select number of (mostly English) tasks, the availability of labeled data is sparse or non-existent for the vast majority of use-cases. To alleviate this, unsupervised learning and a wide array of data augmentation techniques have been developed (Hedderich et al., 2021a). However, unsupervised learning often requires massive amounts of unlabeled data and also fails to perform in difficult (low-resource) data settings, i.e., if there is an increased distance between the source and target data distributions (Kim et al., 2020). This distributional distance can be the case if there is a domain drift or large linguistic distance between the...
With the advent of deep neural networks in recent years, Neural Machine Translation (NMT) systems ha...
This paper describes the methods behind the systems submitted by the University of Groningen for the...
This paper describes the methods behind the systems submitted by the University of Groningen for the...
Unsupervised Machine Translation hasbeen advancing our ability to translatewithout parallel data, bu...
In the context of neural machine translation, data augmentation (DA) techniques may be used for gene...
Neural machine translation (NMT), where neural networks are used to generate translations, has revol...
192 p.Modern machine translation relies on strong supervision in the form of parallel corpora. Such ...
Self-Supervised Learning (SSL) using huge unlabeled data has been successfully explored for image an...
There has been an explosion in unstructured text data in recent years with services like Twitter, Fa...
In recent years, Neural Machine Translation (NMT) has been shown to be more effective than phrase-ba...
Neural Machine Translation (NMT) models are typically trained by considering humans as end-users and...
Recent advances in Neural Machine Translation (NMT) systems have achieved impressive results on lang...
International audienceSelf-Supervised Learning (SSL) using huge unlabeled data has been successfully...
There are several approaches for improving neural machine translation for low-resource languages: mo...
The goal of neural machine translation (NMT) is to build an end-to-end system that automatically tr...
With the advent of deep neural networks in recent years, Neural Machine Translation (NMT) systems ha...
This paper describes the methods behind the systems submitted by the University of Groningen for the...
This paper describes the methods behind the systems submitted by the University of Groningen for the...
Unsupervised Machine Translation hasbeen advancing our ability to translatewithout parallel data, bu...
In the context of neural machine translation, data augmentation (DA) techniques may be used for gene...
Neural machine translation (NMT), where neural networks are used to generate translations, has revol...
192 p.Modern machine translation relies on strong supervision in the form of parallel corpora. Such ...
Self-Supervised Learning (SSL) using huge unlabeled data has been successfully explored for image an...
There has been an explosion in unstructured text data in recent years with services like Twitter, Fa...
In recent years, Neural Machine Translation (NMT) has been shown to be more effective than phrase-ba...
Neural Machine Translation (NMT) models are typically trained by considering humans as end-users and...
Recent advances in Neural Machine Translation (NMT) systems have achieved impressive results on lang...
International audienceSelf-Supervised Learning (SSL) using huge unlabeled data has been successfully...
There are several approaches for improving neural machine translation for low-resource languages: mo...
The goal of neural machine translation (NMT) is to build an end-to-end system that automatically tr...
With the advent of deep neural networks in recent years, Neural Machine Translation (NMT) systems ha...
This paper describes the methods behind the systems submitted by the University of Groningen for the...
This paper describes the methods behind the systems submitted by the University of Groningen for the...