Learning a Reversible Embedding Mapping using Bi-Directional Manifold Alignment
This addresses the need for more efficient and reversible cross-lingual embeddings, though it appears incremental as it builds on existing manifold alignment techniques.
The paper tackles the problem of learning reversible embeddings between languages by proposing Bi-Directional Manifold Alignment (BDMA), which reduces the number of models by 50% and achieves equivalent or better performance compared to standard unidirectional methods.
We propose a Bi-Directional Manifold Alignment (BDMA) that learns a non-linear mapping between two manifolds by explicitly training it to be bijective. We demonstrate BDMA by training a model for a pair of languages rather than individual, directed source and target combinations, reducing the number of models by 50%. We show that models trained with BDMA in the "forward" (source to target) direction can successfully map words in the "reverse" (target to source) direction, yielding equivalent (or better) performance to standard unidirectional translation models where the source and target language is flipped. We also show how BDMA reduces the overall size of the model.