The IIT Bombay Hindi-English Translation System at WMT 2014

Dungarwal, Piyush; Chatterjee, Rajen; Mishra, Abhijit; Kunchukuttan, Anoop; Shah, Ritesh; Bhattacharyya, Pushpak

doi:10.3115/v1/w14-3308

Cited by 26 publications

(17 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…AFRL, AFRL-PE Air Force Research Lab (Schwartz et al, 2014) CIMS University of Stuttgart / University of Munich CMU Carnegie Mellon University (Matthews et al, 2014) CU-* Charles University, Prague DCU-FDA Dublin City University DCU-ICTCAS Dublin City University (Li et al, 2014b) DCU-LINGO24 Dublin City University / Lingo24 (wu et al, 2014) EU-BRIDGE EU-BRIDGE Project KIT Karlsruhe Institute of Technology IIT-BOMBAY IIT Bombay (Dungarwal et al, 2014) IIIT-HYDERABAD IIIT Hyderabad…”

Section: Id Institutionmentioning

confidence: 99%

Findings of the 2014 Workshop on Statistical Machine Translation

Bojar

Buck

Federmann

et al. 2014

Proceedings of the Ninth Workshop on Statistical Machine Translation

555

543

View full text Add to dashboard Cite

This paper presents the results of the WMT14 shared tasks, which included a standard news translation task, a separate medical translation task, a task for run-time estimation of machine translation quality, and a metrics task. This year, 143 machine translation systems from 23 institutions were submitted to the ten translation directions in the standard translation task. An additional 6 anonymized systems were included, and were then evaluated both automatically and manually. The quality estimation task had four subtasks, with a total of 10 teams, submitting 57 entries.

show abstract

Section: Id Institutionmentioning

confidence: 99%

Findings of the 2014 Workshop on Statistical Machine Translation

Bojar

Buck

Federmann

et al. 2014

Proceedings of the Ninth Workshop on Statistical Machine Translation

555

543

View full text Add to dashboard Cite

show abstract

“…The results further improve on using Bi-directional LSTM with attention to give a BLEU score of 9.23, comparable to (Dungarwal et al, 2014), a system fully trained on the WMT training corpus.…”

Section: Experiments and Resultsmentioning

confidence: 93%

Building an Effective MT System for English-Hindi Using RNN's

Agrawal¹,

Sharma²

2017

IJAIA

View full text Add to dashboard Cite

1.INTRODUCTIONDeep learning is a rapidly advancing approach to machine learning and has shown promising performance when applied to a variety of tasks like image recognition, speech processing, natural language processing, cognitive modelling and so on. Deep Learning involves using large neural networks for training a model for a specific task. This paper demonstrates the application of deep learning for Machine Translation of English ! Hindi, two linguistically distant and widely spoken languages. The application of deep neural networks to Machine Translation has been demonstrated by (Kalchbrenner and Blunsom, 2013;Sutskever et al., 2014;Cho et al., 2014;Bahdanau et al., 2014) and it has shown promising results for various language pairs.In this paper, we experiment with different deep learning architectures. These include Gated Recurrent Units (GRUs), Long Short Term Memory Units (LSTMs) and addition of attention mechanism to each of these architectures. We demonstrate that the best performance for English -> Hindi MT is generally obtained using Bi-directional LSTMs with attention mechanism and in some cases with GRUs with attention mechanism. The Bi-directional LSTMs generally show better performance for compound sentences and larger context windows.We show manual samples of output translations and provide their evaluation to demonstrate the effectiveness of different architectures. We describe the motivation behind the choice of RNNs in detail in Section 3. We briefly discuss related work in Section 2, followed by the description of our neural network model in Section 4. The experiments and results are discussed in Section 5. The paper is concluded in Section 6.

show abstract

“…The re-ranking task in machine translation can be defined as re-scoring the n-best list of translations, wherein a number of language models are deployed along with features of source or target language. (Dungarwal et al, 2014) described the benefits of re-ranking the translation hypothesis using simple n-gram based language model. In recent years, the use of RNNLM have shown significant improvements over the traditional n-gram models (Sundermeyer et al, 2013).…”

Section: Related Workmentioning

confidence: 99%

Reducing the Impact of Data Sparsity in Statistical Machine Translation

Singla

Sachdeva

Bangalore

et al. 2014

Proceedings of SSST-8, Eighth Workshop on Syntax, Semantics and Structure in Statistical Translation

View full text Add to dashboard Cite

Morphologically rich languages generally require large amounts of parallel data to adequately estimate parameters in a statistical Machine Translation(SMT) system. However, it is time consuming and expensive to create large collections of parallel data. In this paper, we explore two strategies for circumventing sparsity caused by lack of large parallel corpora. First, we explore the use of distributed representations in an Recurrent Neural Network based language model with different morphological features and second, we explore the use of lexical resources such as WordNet to overcome sparsity of content words.

show abstract

The IIT Bombay Hindi-English Translation System at WMT 2014

Cited by 26 publications

References 12 publications

Findings of the 2014 Workshop on Statistical Machine Translation

Findings of the 2014 Workshop on Statistical Machine Translation

Building an Effective MT System for English-Hindi Using RNN's

Reducing the Impact of Data Sparsity in Statistical Machine Translation

Contact Info

Product

Resources

About