Powerful sentence encoders trained for multiple languages are on the rise. These systems are capable of embedding a wide range of linguistic properties into vector representations. While explicit probing tasks can be used to verify the presence of specific linguistic properties, it is unclear whether the vector representations can be manipulated to indirectly steer such properties. For efficient learning, we investigate the use of a geometric mapping in embedding space to transform linguistic properties, without any tuning of the pre-trained sentence encoder or decoder. We validate our approach on three linguistic properties using a pre-trained multilingual autoencoder and analyze the results in both monolingual and cross-lingual settings.
@article{arxiv.2104.03630,
title = {A Simple Geometric Method for Cross-Lingual Linguistic Transformations with Pre-trained Autoencoders},
author = {Maarten De Raedt and Fréderic Godin and Pieter Buteneers and Chris Develder and Thomas Demeester},
journal= {arXiv preprint arXiv:2104.03630},
year = {2021}
}