Microsoft Research Asia (MSRA) has achieved eight of the top places available in a machine translation challenge organised by the 2019 Conference on Machine Translation (WMT19), out of 11 tasks undertaken. There are 19 machine translation categories at WMT this year.
For Asian languages, MSRA achieved first place in machine translation tasks for Chinese-to-English. Three other tasks placed second in their categories, which included English-to-Kazakh.
As one of the leading machine translation competitions globally, WMT is where researchers demonstrate solutions, as well as further understand the evolution of machine translation. Now in its 14th year, more than 50 teams globally from technology companies, academic institutions and universities participated to demonstrate machine translation capabilities.
"This year, the MSRA team applied innovative algorithms to its system, which significantly improved the quality of the machine translation results. These algorithms were used to improve the platform's learning mechanism, pre-training, network architecture optimisation, data enhancement and other processes required so that the system can perform better," said Tie-Yan Liu, Assistant MD, MSRA.
The algorithms leveraged this year include:
- MADL, multi-agent dual learning
- MASS, masked sequence-to-sequence pre-training
- NAO, automatic neural architecture optimisation
- SCA, soft contextual data augmentation
The achievement follows a 2018 breakthrough whereby researchers in MSRA and at Microsoft Research labs in the US labs reached human parity on a commonly-used test set of news stories, called newstest2017. The system is able to translate sentences from news articles in Chinese into English with the same quality and accuracy as a human.
"The realm of machine translation will continue to evolve with better algorithms, data set and technology. However, much of our research today is really inspired by how we humans do things. Language is complex and nuanced, as people can use different words to express the exact same concept. Hence, developing multidimensional algorithms is important in evolving machine translation systems so that they can deliver better outcomes," said Liu.
"Our achievement at WMT19 serves to the further development of the field, whereby we hope that machine translation can become better in the years to come."
The company's research innovations are already in existing Microsoft solutions. Microsoft Translator, a multilingual machine translation cloud service, has integrated some of the previous solutions developed by Microsoft Research teams globally. The research teams plan to integrate the new algorithms used for this year's WMT challenge to improve the offering.
For Asian languages, MSRA achieved first place in machine translation tasks for Chinese-to-English. Three other tasks placed second in their categories, which included English-to-Kazakh.
As one of the leading machine translation competitions globally, WMT is where researchers demonstrate solutions, as well as further understand the evolution of machine translation. Now in its 14th year, more than 50 teams globally from technology companies, academic institutions and universities participated to demonstrate machine translation capabilities.
"This year, the MSRA team applied innovative algorithms to its system, which significantly improved the quality of the machine translation results. These algorithms were used to improve the platform's learning mechanism, pre-training, network architecture optimisation, data enhancement and other processes required so that the system can perform better," said Tie-Yan Liu, Assistant MD, MSRA.
The algorithms leveraged this year include:
- MADL, multi-agent dual learning
- MASS, masked sequence-to-sequence pre-training
- NAO, automatic neural architecture optimisation
- SCA, soft contextual data augmentation
The achievement follows a 2018 breakthrough whereby researchers in MSRA and at Microsoft Research labs in the US labs reached human parity on a commonly-used test set of news stories, called newstest2017. The system is able to translate sentences from news articles in Chinese into English with the same quality and accuracy as a human.
"The realm of machine translation will continue to evolve with better algorithms, data set and technology. However, much of our research today is really inspired by how we humans do things. Language is complex and nuanced, as people can use different words to express the exact same concept. Hence, developing multidimensional algorithms is important in evolving machine translation systems so that they can deliver better outcomes," said Liu.
"Our achievement at WMT19 serves to the further development of the field, whereby we hope that machine translation can become better in the years to come."
The company's research innovations are already in existing Microsoft solutions. Microsoft Translator, a multilingual machine translation cloud service, has integrated some of the previous solutions developed by Microsoft Research teams globally. The research teams plan to integrate the new algorithms used for this year's WMT challenge to improve the offering.
No comments:
Post a Comment