arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1909.09237cs.CL
收藏

提升互信息的变分神经机器翻译算法

Improved Variational Neural Machine Translation by Promoting Mutual Information

Johns Hopkins University(约翰斯·霍普金斯大学);Facebook(脸书)

Arya D. McCarthy, Xian Li, Jiatao Gu, Ning Dong

摘要:

后崩溃困扰着文本的VAE,特别是对于具有强自回归解码器的条件文本生成。在这项工作中,我们解决了这个问题,变分神经机器翻译显式促进之间的潜在变量和数据的互信息。我们的模型扩展了条件变分自编码器(CVAE)与两个新的成分:首先,我们提出了一个修改的证据下限(ELBO)的目标,明确促进互信息;其次,我们正则化的解码器的概率混合辅助因式分解分布,这是直接预测的潜变量。我们提出的经验结果的Transformer架构,并显示所提出的模型有效地解决后崩溃:潜在变量不再被忽略,在强大的解码器的存在。因此,该模型提高了翻译质量,同时在数据效率和鲁棒性方面表现出卓越的性能。

英文摘要:

Posterior collapse plagues VAEs for text, especially for conditional text generation with strong autoregressive decoders. In this work, we address this problem in variational neural machine translation by explicitly promoting mutual information between the latent variables and the data. Our model extends the conditional variational autoencoder (CVAE) with two new ingredients: first, we propose a modified evidence lower bound (ELBO) objective which explicitly promotes mutual information; second, we regularize the probabilities of the decoder by mixing an auxiliary factorized distribution which is directly predicted by the latent variables. We present empirical results on the Transformer architecture and show the proposed model effectively addressed posterior collapse: latent variables are no longer ignored in the presence of powerful decoder. As a result, the proposed model yields improved translation quality while demonstrating superior performance in terms of data efficiency and robustness.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑