Papers by Željko Agić
JW300: A Wide-Coverage Parallel Corpus for Low-Resource Languages (P19-1)
Copied to clipboard
| Challenge: | a shortage of parallel data in low-resource languages creates a bottleneck for cross-lingual transfer . a massive collection of parallel texts for over 300 diverse languages is our main contribution . |
| Approach: | They propose a parallel corpus of over 300 languages with 100 thousand parallel sentences per language pair on average. |
| Outcome: | The proposed dataset can be used to build cross-lingual word embeddings and multi-source part-of-speech projections. |
MultiQT: Multimodal learning for real-time question tracking in speech (2020.acl-main)
Copied to clipboard
Jakob D. Havtorn, Jan Latko, Joakim Edin, Lars Maaløe, Lasse Borgholt, Lorenzo Belgrano, Nicolai Jacobsen, Regitze Sdun, Željko Agić
| Challenge: | a novel multimodal approach to real-time sequence labeling in speech is proposed . the model treats speech and its own textual representation as two separate modalities . |
| Approach: | They propose a multimodal approach to real-time sequence labeling in speech . they use audio and transcription to jointly learn from a phone call . results show similar pattern of improvements with multimodal learning . |
| Outcome: | The proposed model shows significant gains under adverse noise and limited training data compared to text or audio only under adverse conditions and generalizes to medical symptoms detection. |
Distant Supervision from Disparate Sources for Low-Resource Part-of-Speech Tagging (D18-1)
Copied to clipboard
| Challenge: | Low-resource languages lack manual annotated data to learn basic models such as part-of-speech (POS) taggers. |
| Approach: | They propose a cross-lingual neural part-of-speech tagger that learns from disparate sources of distant supervision in a uniform framework. |
| Outcome: | The proposed model scales to hundreds of low-resource languages without access to gold annotated data. |
Towards Instance-Level Parser Selection for Cross-Lingual Transfer of Dependency Parsers (2020.coling-main)
Copied to clipboard
| Challenge: | Existing methods of cross-lingual parser transfer focus on predicting the best parsers for a low-resource target language globally. |
| Approach: | They propose a cross-lingual parser transfer paradigm that uses instance-level parsers to predict the best parsing for a target language at treebank level. |
| Outcome: | The proposed model outperforms existing models on 13/20 and 14/20 test languages. |
Baselines and Test Data for Cross-Lingual Inference (L18-1)
Copied to clipboard
| Challenge: | Recent research on textual entailment is limited to English, but it is expanding to other languages. |
| Approach: | They propose to extend the research in SNLI-style natural language inference toward multilingual evaluation by using cross-lingual word embeddings and machine translation. |
| Outcome: | The proposed system scores an average accuracy of just over 75%, but it is not perfect. |