ISI Directory

Meryem M'Hamdi

Research Assistant

Research Summary

My main research interest is to adapt cross-lingual unsupervised or supervised approaches to different downstream applications including event extraction, dialogue systems, and studying their probing capabilities at large.  My aim is to devise novel approaches for learning and tuning universal multilingual representations to different domains of applications and studying/increasing their generalization especially in a low resource setting, zero-shot, and few-shot learning scenarios while being evaluated with different flavors of transfer learning. I have been leveraging techniques from multi-task learning/meta-learning to improve the alignment towards a common cross-lingual space.