arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 1802.06901cs.LGcs.CLstat.ML
收藏

确定性非自回归神经元序列的迭代精化建模

Deterministic Non-Autoregressive Neural Sequence Modeling by Iterative Refinement

New York University(纽约大学)

Jason Lee, Elman Mansimov, Kyunghyun Cho

摘要:

提出了一种基于迭代精化的条件非自回归神经序列模型。该模型基于隐变量模型和去噪自编码器的原理设计,一般适用于任何序列生成任务。我们广泛评估了所提出的模型对机器翻译(En-De和En-Ro)和图像字幕生成,并观察到它显着加快解码,同时保持生成质量相媲美的自回归对应。

英文摘要:

We propose a conditional non-autoregressive neural sequence model based on iterative refinement. The proposed model is designed based on the principles of latent variable models and denoising autoencoders, and is generally applicable to any sequence generation task. We extensively evaluate the proposed model on machine translation (En-De and En-Ro) and image caption generation, and observe that it significantly speeds up decoding while maintaining the generation quality comparable to the autoregressive counterpart.

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。

↑