Corpus ID: 11591887

Offline bilingual word vectors, orthogonal transformations and the inverted softmax

@article{Smith2017OfflineBW,
  title={Offline bilingual word vectors, orthogonal transformations and the inverted softmax},
  author={S. L. Smith and David H. P. Turban and S. Hamblin and N. Hammerla},
  journal={ArXiv},
  year={2017},
  volume={abs/1702.03859}
}

• S. L. Smith, David H. P. Turban, +1 author N. Hammerla
• Published 2017
• Computer Science
• ArXiv

Usually bilingual word vectors are trained "online". Mikolov et al. showed they can also be found "offline", whereby two pre-trained embeddings are aligned with a linear transformation, using dictionaries compiled from expert knowledge. In this work, we prove that the linear transformation between two spaces should be orthogonal. This transformation can be obtained using the singular value decomposition. We introduce a novel "inverted softmax" for identifying translation pairs, with which we… Expand

View PDF on arXiv

Save to Library

Create Alert

Cite

Launch Research Feed