Transforming machine translation: a deep learning system reaches news translation quality comparable to human professionals

Popel, Martin; Tomková, Markéta; Tomek, Jakub; Kaiser, Łukasz; Uszkoreit, Jakob; Bojar, Ondřej; Žabokrtský, Zdeněk

doi:10.1038/s41467-020-18073-9

Cited by 210 publications

(133 citation statements)

References 25 publications

Supporting

Mentioning

131

Contrasting

Unclassified

Order By: Relevance

“…Deep Learning systems 18 , have become the state-of-the-art for a wide variety of application domains, including vision (e.g., image classification) and speech (e.g., voice recognition and generation). In language-related tasks (e.g machine translation), these approaches have surpassed human level performance 37 . Deep learning has proven beneficial in numerous NLP tasks, including predicting the next word (language modeling), tagging tasks such as part of speech tags, entities in a sentence (entity recognition), and dependency parsing.…”

Section: Discussion and Future Workmentioning

confidence: 99%

See 1 more Smart Citation

Building a Best-in-Class Automated De-identification Tool for Electronic Health Records Through Ensemble Learning

Murugadoss

Rajasekharan

Malin

et al. 2020

Preprint

View full text Add to dashboard Cite

The natural language portions of an electronic health record (EHR) communicate critical information about disease and treatment progression. However, the presence of personally identifying information in this data constrains its broad reuse. In the United States, the Health Insurance Portability and Accountability Act of 1996 (HIPAA) provides a de-identification standard for the removal of protected health information (PHI). Despite continuous improvements in methods for the automated detection of PHI over time, the residual identifiers in clinical notes continue to pose significant challenges - often requiring manual validation and correction that is not scalable to generate the amount of data needed for modern machine learning tools. In this paper, we describe an automated de-identification system that employs an ensemble architecture, incorporating attention-based deep learning models and rule based methods, supported by heuristics for detecting PHI in EHR data. Upon detection of PHI, the system transforms these detected identifiers into plausible, though fictional, surrogates to further obfuscate any leaked identifier. We evaluated the system with a publicly available dataset of 515 notes from the I2B2 2014 de-identification challenge and a dataset of 10,000 notes from the Mayo Clinic. We compared our approach with other existing tools considered best-in-class. The results indicated a recall of 0.992 and 0.994 and a precision of 0.979 and 0.967 on the I2B2 and the Mayo Clinic data, respectively.

show abstract

Section: Discussion and Future Workmentioning

confidence: 99%

“…The I2B2 2014 De-identification and Heart Disease Risk Factors challenge [29] is a publicly available dataset of clinical documents with annotated PHI elements. This dataset consists of a training set of 792 clinical notes and a test set of 515 clinical notes.…”

Section: Dataset Description I2b2 Dataset Descriptionmentioning

confidence: 99%

Building a Best-in-Class Automated De-identification Tool for Electronic Health Records Through Ensemble Learning

Murugadoss

Rajasekharan

Malin

et al. 2020

Preprint

View full text Add to dashboard Cite

show abstract

“…In order to verify the application effect of a business English translation framework based on speech recognition and wireless communication, the simulation experiment is carried out. In the experiment, design a translation system based on deep learning [7], an interactive English-Chinese translation system based on feature extraction algorithm [8], and a system based on joint minimum Bayesian fusion [9] as a comparison.…”

Section: Methodsmentioning

confidence: 99%

“…erefore, a program to automatically calculate the average value was written in Java language. e error correction results of the designed system, the system in [7], the system in [8], and the system in [9] were counted.…”

Section: Comparison Of Error Correction Rate and Total Errormentioning

confidence: 99%

“…erefore, it is of great significance to study a multifunctional business English translation tool. e authors of [7] designed a translation system based on deep learning and proposes a deep learning system for context-aware blind assessment. It is superior to traditional English translation tools in retaining the meaning of the text (translation adequacy), and the fluency of the translation system has been significantly improved.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Research on Business English Translation Framework Based on Speech Recognition and Wireless Communication

Wu²

2021

Mobile Information Systems

View full text Add to dashboard Cite

In order to improve the accuracy of English translation, reduce the error rate of translation results, and increase the correction rate of translation, this paper proposes a business English translation architecture design based on speech recognition and wireless communication. The architecture is partitioned according to the functions of the overall system design, and the voice acquisition module, voice processing module, and peripheral circuit module are designed according to functional requirements. Among them, speech recognition helps users to perform language translation to reduce the possibility of errors in the translation process. At the same time, it uses wireless communication technology to construct a business English translation corpus to meet the personal needs of users. The paper also uses an improved translation model for translation error correction and intelligent proofreading, which improves the reliability of translation results. Experimental results show that the system has a high error correction rate and error correction rate, and the translation results have a certain degree of reliability, which fully verifies the effectiveness and application value of the system.

show abstract

Toward automatic generation of control structures for process flow diagrams with large language models

Hirtreiter,

Schulze Balhorn,

Schweidtmann

2023

AIChE Journal

View full text Add to dashboard Cite

Developing Piping and Instrumentation Diagrams (P&IDs) is a crucial step during process development. We propose a data‐driven method for the prediction of control structures. Our methodology is inspired by end‐to‐end transformer‐based human language translation models. We cast the control structure prediction as a translation task where Process Flow Diagrams (PFDs) without control structures are translated to PFDs with control structures. We represent the topology of PFDs as strings using the SFILES 2.0 notation. We pretrain our model using generated PFDs to learn the grammatical structure. Thereafter, the model is fine‐tuned leveraging transfer learning on real PFDs. The model achieved a top‐5 accuracy of 74.8% on 10,000 generated PFDs and 89.2% on 100,000 generated PFDs. These promising results show great potential for AI‐assisted process engineering. The tests on a dataset of 312 real PFDs indicate the need for a larger PFD dataset for industry applications and hybrid artificial intelligence solutions.

show abstract

Transforming machine translation: a deep learning system reaches news translation quality comparable to human professionals

Cited by 210 publications

References 25 publications

Building a Best-in-Class Automated De-identification Tool for Electronic Health Records Through Ensemble Learning

Building a Best-in-Class Automated De-identification Tool for Electronic Health Records Through Ensemble Learning

Research on Business English Translation Framework Based on Speech Recognition and Wireless Communication

Toward automatic generation of control structures for process flow diagrams with large language models

Contact Info

Product

Resources

About