Bilingual-GAN: A Step Towards Parallel Text Generation

TL;DR

This paper introduces Bilingual-GAN, a novel model that combines latent space and adversarial training to generate parallel bilingual sentences and perform bidirectional translation.

Contribution

It proposes a new approach that leverages shared latent space and GANs for simultaneous bilingual text generation and translation, integrating autoencoders and back-translation.

Findings

Effective bilingual sentence generation demonstrated on Europarl and Multi30k datasets.

Achieved competitive results in supervised and unsupervised translation tasks.

Introduced a shared latent space approach for bidirectional translation.

Abstract

Latent space based GAN methods and attention based sequence to sequence models have achieved impressive results in text generation and unsupervised machine translation respectively. Leveraging the two domains, we propose an adversarial latent space based model capable of generating parallel sentences in two languages concurrently and translating bidirectionally. The bilingual generation goal is achieved by sampling from the latent space that is shared between both languages. First two denoising autoencoders are trained, with shared encoders and back-translation to enforce a shared latent state between the two languages. The decoder is shared for the two translation directions. Next, a GAN is trained to generate synthetic "code" mimicking the languages' shared latent space. This code is then fed into the decoder to generate text in either language. We perform our experiments on Europarl and Multi30k datasets, on the English-French language pair, and document our performance using both supervised and unsupervised machine translation.