search

viking-sudo-rm / voynich2vec

Applying word2vec embeddings to the problem of deciphering the Voynich manuscript.
7 stars 0 forks source link

Non-Adversarial Unsupervised Word Translation #12

Open viking-sudo-rm opened 6 years ago

viking-sudo-rm commented 6 years ago

https://arxiv.org/abs/1801.06126

This method does the same thing as MUSE, but uses a non-adversarial training method. This means it is much easier to train and less data-hungry. Therefore, it might not fail as disastrously on Voynich.

chirila commented 6 years ago

cool. Is code available?

On Thu, Oct 4, 2018 at 4:54 PM Will Merrill notifications@github.com wrote:

https://arxiv.org/abs/1801.06126

This method does the same thing as MUSE, but uses a non-adversarial training method. This means it is much easier to train and less data-hungry. Therefore, it might not fail as disastrously on Voynich.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/viking-sudo-rm/voynich2vec/issues/12, or mute the thread https://github.com/notifications/unsubscribe-auth/AP8oR58-XwN3SB2wtKB542qBP83RI0Yjks5uhnWcgaJpZM4XI5Qr .

--

Claire Bowern Professor, Director of Graduate Studies Chair: Yale Women Faculty Forum (wff.yale.edu) Department of Linguistics New Haven, CT 06511

viking-sudo-rm commented 6 years ago

Yes. There are actually several papers related to this.

A particularly nice one (with code) is: https://arxiv.org/pdf/1809.03633.pdf https://github.com/xrc10/unsup-cross-lingual-embedding-transfer