Edge NLP for Efficient Machine Translation in Low Connectivity Areas

Tess Watt; Christos Chrysoulas; Dimitra Gkatzia

doi:10.1109/WF-IoT58464.2023.10539577

Edge NLP for Efficient Machine Translation in Low Connectivity Areas

Tess Watt, Christos Chrysoulas, Dimitra Gkatzia

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

Machine translation (MT) usually requires connectivity and access to the cloud which is often limited in many parts of the world, including hard to reach rural areas. Natural language processing (NLP) on the edge aims to solve this problem by processing language data closer to the source. To achieve this, 100 sentence pairs were stored and processed on a Raspberry Pi, and a recurrent neural network (RNN) using the long short-term memory (LSTM) architecture was used for MT. We are focusing on translating between English and Hausa, a low-resource language spoken in West Africa. It was found that the developed prototype produced 'good and fluent translations' with a training accuracy of 91%. The model also achieved a BLEU score of 73.5, compared to the existing models that have scores of 22.2 and below.

Original language	English
Title of host publication	9th IEEE World Forum on Internet of Things (WF-IoT)
Publisher	IEEE
ISBN (Electronic)	9798350311617
DOIs	https://doi.org/10.1109/WF-IoT58464.2023.10539577
Publication status	Published - 30 May 2024
Event	9th IEEE World Forum on Internet of Things 2023 - Hybrid, Aveiro, Portugal Duration: 12 Oct 2023 → 27 Oct 2023

Conference

Conference	9th IEEE World Forum on Internet of Things 2023
Abbreviated title	WF-IoT 2023
Country/Territory	Portugal
City	Aveiro
Period	12/10/23 → 27/10/23

Keywords

artificial intelligence
computation offloading
edge computing
machine learning

ASJC Scopus subject areas

Artificial Intelligence
Computer Networks and Communications
Hardware and Architecture
Safety, Risk, Reliability and Quality
Modelling and Simulation
Instrumentation

Access to Document

10.1109/WF-IoT58464.2023.10539577

Cite this

@inproceedings{b1ce777d3c374e90998b4e4f2c840c62,

title = "Edge NLP for Efficient Machine Translation in Low Connectivity Areas",

abstract = "Machine translation (MT) usually requires connectivity and access to the cloud which is often limited in many parts of the world, including hard to reach rural areas. Natural language processing (NLP) on the edge aims to solve this problem by processing language data closer to the source. To achieve this, 100 sentence pairs were stored and processed on a Raspberry Pi, and a recurrent neural network (RNN) using the long short-term memory (LSTM) architecture was used for MT. We are focusing on translating between English and Hausa, a low-resource language spoken in West Africa. It was found that the developed prototype produced 'good and fluent translations' with a training accuracy of 91%. The model also achieved a BLEU score of 73.5, compared to the existing models that have scores of 22.2 and below.",

keywords = "artificial intelligence, computation offloading, edge computing, machine learning",

author = "Tess Watt and Christos Chrysoulas and Dimitra Gkatzia",

note = "Publisher Copyright: {\textcopyright} 2023 IEEE.; 9th IEEE World Forum on Internet of Things 2023, WF-IoT 2023 ; Conference date: 12-10-2023 Through 27-10-2023",

year = "2024",

month = may,

day = "30",

doi = "10.1109/WF-IoT58464.2023.10539577",

language = "English",

booktitle = "9th IEEE World Forum on Internet of Things (WF-IoT)",

publisher = "IEEE",

address = "United States",

}

TY - GEN

T1 - Edge NLP for Efficient Machine Translation in Low Connectivity Areas

AU - Watt, Tess

AU - Chrysoulas, Christos

AU - Gkatzia, Dimitra

PY - 2024/5/30

Y1 - 2024/5/30

N2 - Machine translation (MT) usually requires connectivity and access to the cloud which is often limited in many parts of the world, including hard to reach rural areas. Natural language processing (NLP) on the edge aims to solve this problem by processing language data closer to the source. To achieve this, 100 sentence pairs were stored and processed on a Raspberry Pi, and a recurrent neural network (RNN) using the long short-term memory (LSTM) architecture was used for MT. We are focusing on translating between English and Hausa, a low-resource language spoken in West Africa. It was found that the developed prototype produced 'good and fluent translations' with a training accuracy of 91%. The model also achieved a BLEU score of 73.5, compared to the existing models that have scores of 22.2 and below.

AB - Machine translation (MT) usually requires connectivity and access to the cloud which is often limited in many parts of the world, including hard to reach rural areas. Natural language processing (NLP) on the edge aims to solve this problem by processing language data closer to the source. To achieve this, 100 sentence pairs were stored and processed on a Raspberry Pi, and a recurrent neural network (RNN) using the long short-term memory (LSTM) architecture was used for MT. We are focusing on translating between English and Hausa, a low-resource language spoken in West Africa. It was found that the developed prototype produced 'good and fluent translations' with a training accuracy of 91%. The model also achieved a BLEU score of 73.5, compared to the existing models that have scores of 22.2 and below.

KW - artificial intelligence

KW - computation offloading

KW - edge computing

KW - machine learning

UR - http://www.scopus.com/inward/record.url?scp=85195392513&partnerID=8YFLogxK

U2 - 10.1109/WF-IoT58464.2023.10539577

DO - 10.1109/WF-IoT58464.2023.10539577

M3 - Conference contribution

AN - SCOPUS:85195392513

BT - 9th IEEE World Forum on Internet of Things (WF-IoT)

PB - IEEE

T2 - 9th IEEE World Forum on Internet of Things 2023

Y2 - 12 October 2023 through 27 October 2023

ER -

Edge NLP for Efficient Machine Translation in Low Connectivity Areas

Abstract

Conference

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this