More specifically, in the workshop the participants will have the chance to explore the current state-of-the-art of cross-lingual and multi-lingual models, like for example LASER and BERT, and acquire hands-on experience on leveraging such approaches in order to solve cross-lingual tasks.
The workshop will be structured as follows. A short introduction to word embeddings and cross-lingual word embeddings will be given, where the participants will be enabled to leverage them in order to solve downstream tasks (e.g. sentiment classification, sentence retrieval etc.) with baseline approaches. Then, the participants will have the chance to apply state-of-the-art neural models that use cross-lingual word embeddings on various tasks like sentiment classification or cross-lingual sentence retrieval.