Unsupervised Cross-Lingual Transfer of Structured Predictors without Source Data
This addresses the problem of providing structured prediction technologies to domains with scarce or private training data, offering an incremental improvement over existing unsupervised transfer methods.
The paper tackles unsupervised cross-lingual transfer of structured predictors without source data by proposing a method that multiplies marginal probabilities of substructures for aggregation, which is shown to be better than prior union-based approaches. Results on 18 languages for dependency parsing and part-of-speech tagging demonstrate less noisy labels and improved performance.
Providing technologies to communities or domains where training data is scarce or protected e.g., for privacy reasons, is becoming increasingly important. To that end, we generalise methods for unsupervised transfer from multiple input models for structured prediction. We show that the means of aggregating over the input models is critical, and that multiplying marginal probabilities of substructures to obtain high-probability structures for distant supervision is substantially better than taking the union of such structures over the input models, as done in prior work. Testing on 18 languages, we demonstrate that the method works in a cross-lingual setting, considering both dependency parsing and part-of-speech structured prediction problems. Our analyses show that the proposed method produces less noisy labels for the distant supervision.