A Multi-task Approach to Learning Multilingual Representations
Open accessThe DOI matches its Crossref record; 38 of the 38 references it cites were found in a registry or an index.
Abstract
We present a novel multi-task modeling approach to learning multilingual distributed representations of text.Our system learns word and sentence embeddings jointly by training a multilingual skipgram model together with a cross-lingual sentence similarity model.Our architecture can transparently use both monolingual and sentence aligned bilingual corpora to learn multilingual embeddings, thus covering a vocabulary significantly larger than the vocabulary of the bilingual corpora alone.Our model shows competitive performance in a standard crosslingual document classification task.We also show the effectiveness of our method in a limited resource scenario.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The check wasn’t confident either way, so it gives no verdict — never disputed.
In the citing paper
“Besides, Hewitt and Manning [125], Jawahar et al. [126], Kim et al. [127] proposed several methods to extract dependency trees and constituency trees from BERT, which proved the BERT's ability to encode syntax structure.”
From Pre-trained models for natural language processing: A survey (2020) · cites this paper as [127]
Experimental — accuracy not yet measured
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Hochreiter, S., & Schmidhuber, J. (1997) · Neural Computation · doi:10.1162/neco.1997.9.8.1735 · Cited by 102,318 · cited 1×
“Cross-lingual Sentence Similarity: We process word embeddings with a bi-directional LSTM (Hochreiter et al., 2001; Hochreiter and Schmidhuber, 1997) and then take an average of the LSTM outputs (Figure 2).”
MethodAdam: A Method for Stochastic Optimization
Kingma, D. P., & Ba, J. (2014) · arXiv · doi:10.48550/arxiv.1412.6980 · Cited by 98,946 · cited 1×
“The model is trained using the Adam optimizer (Kingma and Ba, 2014) with a learning rate of 0.”
MethodGlove: Global Vectors for Word Representation
Pennington, J., Socher, R., & Manning, C. (2014) · Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.3115/v1/d14-1162 · Cited by 34,071 · cited 2×
“Learning distributed representations of text, whether it be at the level of words, phrases, sentences or documents has been one of the most widely researched subjects in natural language processing in recent years (Mikolov et al., 2013; Pennington et al., 2014; Gouws et al., 2015; Socher et al., 2010; Pham et al., 201…”
BackgroundDistributed Representations of Words and Phrases and their Compositionality
Mikolov, T., Sutskever, I., Chen, K., Corrado, G., & Dean, J. (2013) · arXiv · doi:10.48550/arxiv.1310.4546 · Cited by 24,533 · cited 2×
“Learning distributed representations of text, whether it be at the level of words, phrases, sentences or documents has been one of the most widely researched subjects in natural language processing in recent years (Mikolov et al., 2013; Pennington et al., 2014; Gouws et al., 2015; Socher et al., 2010; Pham et al., 201…”
BackgroundNeural Machine Translation by Jointly Learning to Align and Translate
Bahdanau, D., Cho, K., & Bengio, Y. (2014) · arXiv · doi:10.48550/arxiv.1409.0473 · Cited by 16,213 · cited 2×
“Word/sentence/document embeddings, as they are now commonly referred to, have quickly become essential ingredients of larger and more complex NLP systems looking to leverage the rich semantic and linguistic information present in distributed representations (Bengio et al., 2003; Maas et al., 2011; Collobert et al., 20…”
BackgroundLearning Word Vectors for Sentiment Analysis
Maas, A. L., Daly, R. E., Pham, P. T., Huang, D., Ng, A. Y., & Potts, C. (2011) · Cited by 6,960
Natural Language Processing (almost) from Scratch
Collobert, R., Weston, J., Bottou, L., Karlen, M., Kavukcuoglu, K., & Kuksa, P. P. (2011) · arXiv (Cornell University) · doi:10.48550/arxiv.1103.0398 · Cited by 4,962 · cited 2×
“Word/sentence/document embeddings, as they are now commonly referred to, have quickly become essential ingredients of larger and more complex NLP systems looking to leverage the rich semantic and linguistic information present in distributed representations (Bengio et al., 2003; Maas et al., 2011; Collobert et al., 20…”
BackgroundDistributed Representations of Sentences and Documents
Le, Q. V., & Mikolov, T. (2014) · arXiv (Cornell University) · doi:10.48550/arxiv.1405.4053 · Cited by 4,934 · cited 1×
“Learning distributed representations of text, whether it be at the level of words, phrases, sentences or documents has been one of the most widely researched subjects in natural language processing in recent years (Mikolov et al., 2013; Pennington et al., 2014; Gouws et al., 2015; Socher et al., 2010; Pham et al., 201…”
BackgroundHierarchical Attention Networks for Document Classification
Yang, Z., Yang, D., Dyer, C., He, X., Smola, A., & Hovy, E. (2016) · Proceedings of the 2016 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies · doi:10.18653/v1/n16-1174 · Cited by 4,712 · cited 1×
“We observed that this additional layer is beneficial in most settings and this is consistent with the observations of recent works on learning sentence and document embeddings such as (Conneau et al., 2017; Yang et al., 2016)”
ResultA Neural Probabilistic Language Model
Bengio, Y., Ducharme, R., Vincent, P., Jauvin, C., Jaz, K., Hofmann, T., Poggio, T., & Shawe-Taylor, J. S. (2000) · Cited by 3,479
(2005) · Cited by 3,125
Recurrent Neural Network Regularization
Zaremba, W., Sutskever, I., & Vinyals, O. (2014) · arXiv (Cornell University) · doi:10.48550/arxiv.1409.2329 · Cited by 2,123 · cited 1×
“There are various implementations of LSTMs available; in this work we use an implementation based on (Zaremba et al., 2014).”
MethodSupervised Learning of Universal Sentence Representations from Natural Language Inference Data
Conneau, A., Kiela, D., Schwenk, H., Barrault, L., & Bordes, A. (2017) · Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing · doi:10.18653/v1/d17-1070 · Cited by 2,049 · cited 4×
“Learning distributed representations of text, whether it be at the level of words, phrases, sentences or documents has been one of the most widely researched subjects in natural language processing in recent years (Mikolov et al., 2013; Pennington et al., 2014; Gouws et al., 2015; Socher et al., 2010; Pham et al., 201…”
BackgroundResultInfluentialA Fast and Accurate Dependency Parser using Neural Networks
Chen, D., & Manning, C. (2014) · Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.3115/v1/d14-1082 · Cited by 1,795 · cited 1×
“Word/sentence/document embeddings, as they are now commonly referred to, have quickly become essential ingredients of larger and more complex NLP systems looking to leverage the rich semantic and linguistic information present in distributed representations (Bengio et al., 2003; Maas et al., 2011; Collobert et al., 20…”
BackgroundGoogle’s Multilingual Neural Machine Translation System: Enabling Zero-Shot Translation
Johnson, M., Schuster, M., Le, Q. V., Krikun, M., Wu, Y., Chen, Z., Thorat, N., Viégas, F., Wattenberg, M., Corrado, G., et al. (2017) · Transactions of the Association for Computational Linguistics · doi:10.1162/tacl_a_00065 · Cited by 1,747 · cited 2×
“Multilingual embeddings open up the possibility of transferring knowledge across languages and building complex systems even for languages with limited amount of supervised resources (Ammar et al., 2016; Johnson et al., 2016).”
BackgroundKiros, R., Zhu, Y., Salakhutdinov, R., Zemel, R. S., Torralba, A., Urtasun, R., & Fidler, S. (2015) · arXiv · doi:10.48550/arxiv.1506.06726 · Cited by 1,293 · cited 2×
“Learning distributed representations of text, whether it be at the level of words, phrases, sentences or documents has been one of the most widely researched subjects in natural language processing in recent years (Mikolov et al., 2013; Pennington et al., 2014; Gouws et al., 2015; Socher et al., 2010; Pham et al., 201…”
BackgroundGradient Flow in Recurrent Nets: The Difficulty of Learning LongTerm Dependencies
(2009) · A Field Guide to Dynamical Recurrent Networks · doi:10.1109/9780470544037.ch14 · Cited by 1,246
Improving Vector Space Word Representations Using Multilingual Correlation
Faruqui, M., & Dyer, C. (2014) · Proceedings of the 14th Conference of the European Chapter of the Association for Computational Linguistics · doi:10.3115/v1/e14-1049 · Cited by 620 · cited 1×
“Research that has been taking place in the context of distributed text representations is learning multilingual text representations shared across languages (Faruqui and Dyer, 2014; Bengio and Corrado, 2015; Luong et al., 2015).”
BackgroundInducing Crosslingual Distributed Representations of Words
Klementiev, A., Titov, I., & Bhattarai, B. (2012) · International Conference on Computational Linguistics · Cited by 389
Bilingual Word Representations with Monolingual Quality in Mind
Luong, T., Pham, H., & Manning, C. D. (2015) · Proceedings of the 1st Workshop on Vector Space Modeling for Natural Language Processing · doi:10.3115/v1/w15-1521 · Cited by 315 · cited 6×
“2014), BiSkip-UnsupAlign (Luong et al., 2015) and para doc (Pham et al.”
BackgroundMethodInfluentialBilBOWA: Fast Bilingual Distributed Representations without Word Alignments
Gouws, S., Bengio, Y., & Corrado, G. S. (2014) · arXiv (Cornell University) · doi:10.48550/arxiv.1410.2455 · Cited by 285 · cited 2×
“Research that has been taking place in the context of distributed text representations is learning multilingual text representations shared across languages (Faruqui and Dyer, 2014; Bengio and Corrado, 2015; Luong et al., 2015).”
BackgroundMultilingual Models for Compositional Distributed Semantics
Hermann, K. M., & Blunsom, P. (2014) · Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · doi:10.3115/v1/p14-1006 · Cited by 268 · cited 3×
“Without the LSTM layer, this loss is similar to the BiCVM loss (Hermann and Blunsom, 2014) except that we use also the reversed sample (T, S) to train the model, therefore showing each pair of sentences to the model two times per epoch.”
BackgroundMethodResultAmmar, W., Mulcaire, G., Ballesteros, M., Dyer, C., & Smith, N. A. (2016) · Transactions of the Association for Computational Linguistics · doi:10.1162/tacl_a_00109 · Cited by 232 · cited 2×
“Multilingual embeddings open up the possibility of transferring knowledge across languages and building complex systems even for languages with limited amount of supervised resources (Ammar et al., 2016; Johnson et al., 2016).”
BackgroundLearning Joint Multilingual Sentence Representations with Neural Machine Translation
Schwenk, H., & Douze, M. (2017) · Proceedings of the 2nd Workshop on Representation Learning for NLP · doi:10.18653/v1/w17-2619 · Cited by 197 · cited 2×
“, 2014; Hermann and Blunsom, 2014; Mogadala and Rettinger, 2016) or parametric composition models learned along with the word embeddings (Schwenk et al., 2017).”
BackgroundStarSpace: Embed All The Things!
Wu, L., Fisch, A., Chopra, S., Adams, K., Bordes, A., & Weston, J. (2018) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v32i1.11996 · Cited by 187
A survey of cross-lingual embedding models.
Ruder, S. (2017) · Cited by 121
Google's Multilingual Neural Machine Translation System: Enabling Zero-Shot TranslationPreprint
Johnson, M., Schuster, M., Le, Q. V., et al. (2016) · arXiv (Cornell University) · doi:10.48550/arxiv.1611.04558 · Cited by 109 · cited 2×
“Multilingual embeddings open up the possibility of transferring knowledge across languages and building complex systems even for languages with limited amount of supervised resources (Ammar et al., 2016; Johnson et al., 2016).”
BackgroundMultilingual Models for Compositional Distributed SemanticsPreprint
Hermann, K. M., & Blunsom, P. (2014) · arXiv (Cornell University) · doi:10.48550/arxiv.1404.4641 · Cited by 103 · cited 3×
“Without the LSTM layer, this loss is similar to the BiCVM loss (Hermann and Blunsom, 2014) except that we use also the reversed sample (T, S) to train the model, therefore showing each pair of sentences to the model two times per epoch.”
BackgroundMethodResultStarSpace: Embed All The Things!Preprint
Wu, L., Fisch, A., Chopra, S., et al. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1709.03856 · Cited by 92
Efficient Vector Representation for Documents through Corruption
Chen, M. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1707.02377 · Cited by 84 · cited 1×
“Learning distributed representations of text, whether it be at the level of words, phrases, sentences or documents has been one of the most widely researched subjects in natural language processing in recent years (Mikolov et al., 2013; Pennington et al., 2014; Gouws et al., 2015; Socher et al., 2010; Pham et al., 201…”
BackgroundMogadala, A., & Rettinger, A. (2016) · Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT) · doi:10.18653/v1/n16-1083 · Cited by 80 · cited 1×
“dings (Lauly et al., 2014; Hermann and Blunsom, 2014; Mogadala and Rettinger, 2016) or parametric composition models learned along with the word embeddings (Schwenk et al.”
BackgroundLearning Distributed Representations for Multilingual Text Sequences
Pham, H., Luong, M.-T., & Manning, C. D. (2015) · Workshop on Vector Space Modeling for Natural Language Processing (VSM-NLP) · doi:10.3115/v1/w15-1512 · Cited by 63 · cited 2×
“, 2015) and para doc (Pham et al., 2015a).”
BackgroundJointly optimizing word representations for lexical and sentential tasks with the C-PHRASE model
Pham, N. T., Kruszewski, G., Lazaridou, A., & Baroni, M. (2015) · Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natural Language Processing (Volume 1: Long Papers) · doi:10.3115/v1/p15-1094 · Cited by 46 · cited 1×
“Learning distributed representations of text, whether it be at the level of words, phrases, sentences or documents has been one of the most widely researched subjects in natural language processing in recent years (Mikolov et al., 2013; Pennington et al., 2014; Gouws et al., 2015; Socher et al., 2010; Pham et al., 201…”
BackgroundLearning Multilingual Word Representations using a Bag-of-Words AutoencoderPreprint
Lauly, S., Boulanger, A., & Larochelle, H. (2014) · arXiv (Cornell University) · doi:10.48550/arxiv.1401.1803 · Cited by 41 · cited 2×
“dings (Lauly et al., 2014; Hermann and Blunsom, 2014; Mogadala and Rettinger, 2016) or parametric composition models learned along with the word embeddings (Schwenk et al.”
BackgroundSupervised Learning of Universal Sentence Representations from Natural Language Inference Data
Conneau, A., Kiela, D., Schwenk, H., et al. (2017) · arXiv · doi:10.48550/arxiv.1705.02364 · cited 4×
“Learning distributed representations of text, whether it be at the level of words, phrases, sentences or documents has been one of the most widely researched subjects in natural language processing in recent years (Mikolov et al., 2013; Pennington et al., 2014; Gouws et al., 2015; Socher et al., 2010; Pham et al., 201…”
BackgroundResultInfluentialAmmar, W., Mulcaire, G., Ballesteros, M., Dyer, C., & Smith, N. A. (2016) · arXiv · doi:10.48550/arxiv.1602.01595 · cited 2×
“Multilingual embeddings open up the possibility of transferring knowledge across languages and building complex systems even for languages with limited amount of supervised resources (Ammar et al., 2016; Johnson et al., 2016).”
BackgroundA Survey Of Cross-lingual Word Embedding Models
Ruder, S., Vulić, I., & Søgaard, A. (2017) · arXiv · doi:10.48550/arxiv.1706.04902 · cited 1×
“models, please refer to (Ruder, 2017).”
Learning Joint Multilingual Sentence Representations with Neural Machine Translation
Schwenk, H., & Douze, M. (2017) · arXiv · doi:10.48550/arxiv.1704.04154 · cited 2×
“, 2014; Hermann and Blunsom, 2014; Mogadala and Rettinger, 2016) or parametric composition models learned along with the word embeddings (Schwenk et al., 2017).”
Background
Research works printed but not linked: 3
- Stephan Gouws, Yoshua Bengio, and Greg Corrado. 2015. Bilbowa: Fast bilingual distributed represen- tations without word alignments. In International Conference on Machine Learning, pages 748-756.
- Tomas Mikolov, Ilya Sutskever, Kai Chen, Greg S Cor- rado, and Jeff Dean. 2013. Distributed representa- tions of words and phrases and their compositional- ity. In Advances in neural information processing systems, pages 3111-3119.
- Richard Socher, Christopher D Manning, and An- drew Y Ng. 2010. Learning continuous phrase representations and syntactic parsing with recursive neural networks. In Proceedings of the NIPS-2010 Deep Learning and Unsupervised Feature Learning Workshop, pages 1-9.
Cited by
Pre-trained models for natural language processing: A survey
Qiu, X., Sun, T., Xu, Y., Shao, Y., Dai, N., & Huang, X. (2020) · Science China Technological Sciences · doi:10.1007/s11431-020-1647-3 · Cited by 1,591
“Most of early works focus on learning multilingual word embedding [41, 110, 145], which represents text from multiple languages in a single semantic space.”
BackgroundCross-lingual Transfer Learning for Multilingual Task Oriented Dialog
Schuster, S., Gupta, S., Shah, R., & Lewis, M. (2019) · Proceedings of the 2019 Conference of the North · doi:10.18653/v1/n19-1380 · Cited by 246
“tions using either multilingual MT encoders similar to ours (e.g., Eriguchi et al., 2018; Yu et al., 2018a; Singla et al., 2018) or training encoders and then aligning their vector spaces after pretraining (Conneau et al.”
MethodMulti-Task Learning in Natural Language Processing: An Overview
Chen, S., Zhang, Y., & Yang, Q. (2024) · ACM Computing Surveys · doi:10.1145/3663363 · Cited by 313
“A shallow trunk can be simply the word representation layer [108] while a deep trunk can be the entire model except output layers.”
InfluentialFitzGerald, J., Hench, C., Peris, C., Mackie, S., Rottmann, K., Sanchez, A., Nash, A., Urbach, L., Kakarala, V., Singh, R., et al. (2023) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2023.acl-long.235 · Cited by 369
End-to-End Slot Alignment and Recognition for Cross-Lingual NLU
Xu, W., Haider, B. A., & Mansour, S. (2020) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2020.emnlp-main.410 · Cited by 187
Metadata
- Authors
- Singla, K., Can, D., & Narayanan, S.
- Affiliations
- University of Southern California (US)
- Venue
- Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)
- Published
- 2018
- Pages
- 214-220
- ISSN
- 0736-587X (ISSN-L)
- OpenAlex
- W2798762751
- Cited by
- 22
- References
- 36
- Citations by year
- 2024: 1 · 2023: 3 · 2022: 1 · 2021: 3 · 2020: 5 · 2019: 7 · 2018: 2
- FWCI
- 1.74
- Type
- conference paper
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- cc-by
- Locations
- Proceedings of the conference - Association for Computational Linguistics. Meeting (publishedVersion, cc-by)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
- Physical Sciences › Computer Science › Artificial Intelligence › Natural Language Processing Techniques
- Physical Sciences › Computer Science › Artificial Intelligence › Domain Adaptation and Few-Shot Learning
- Keywords
- multilingual embeddings, sentence embeddings, multi-task learning, skip-gram, bilingual corpus, low-resource languages
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.