Quantifying Linguistic Variation
Add to your list(s)
Download to your calendar using vCal
If you have a question about this talk, please contact Panagiotis Fytas.
Variation in language is ubiquitous, but difficult to define. Starting from Natural Language Processing’s more pragmatic definition that an increase in distributional shift leads to decreases in transferability, we search for quantitative measures of linguistic variation. By identifying subspaces in latent representations which relate to specific syntactic and semantic features, we attempt to disentangle interpretable dimensions of variation. This segmentation allows us to identify suitable data for transfer learning in specialized low-resource scenarios as well as to accurately evaluate the robustness of pre-trained models before fine-tuning them.
This talk is part of the Language Technology Lab Seminars series.
This talk is included in these lists:
Note that ex-directory lists are not shown.
|