arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2212.03232cs.LGcs.AIstat.ML
收藏

使用很少或没有配对数据学习两个序列的联合分布

Learning the joint distribution of two sequences using little or no paired data

Google(谷歌)

Soroosh Mariooryad, Matt Shannon, Siyuan Ma, Tom Bagby, David Kao, Daisy Stanton, Eric Battenberg, RJ Skerry-Ryan

摘要:

我们提出了一个有噪声的通道生成模型的两个序列,例如文本和语音,这使得揭露两种模式之间的关联时,有限的配对数据可用。为了解决一个现实的数据设置下的精确模型的棘手性,我们提出了一个变分推理近似。为了用分类数据训练这个变分模型,我们提出了一个KL编码器丢失方法,它与唤醒-睡眠算法有关。通过仅观察来自边缘的未配对样本来识别联合或条件分布仅在数据分布的某些条件下是可能的,并且我们讨论了可能实现的条件独立性假设的类型,这指导了架构设计。实验结果表明,即使是少量的配对数据(5分钟)也足以在大量未配对数据可用时学习将两种模态(这里的字素和音素)联系起来,从而为在低数据资源机制中采用这种原则性方法铺平了道路。

英文摘要:

We present a noisy channel generative model of two sequences, for example text and speech, which enables uncovering the association between the two modalities when limited paired data is available. To address the intractability of the exact model under a realistic data setup, we propose a variational inference approximation. To train this variational model with categorical data, we propose a KL encoder loss approach which has connections to the wake-sleep algorithm. Identifying the joint or conditional distributions by only observing unpaired samples from the marginals is only possible under certain conditions in the data distribution and we discuss under what type of conditional independence assumptions that might be achieved, which guides the architecture designs. Experimental results show that even tiny amount of paired data (5 minutes) is sufficient to learn to relate the two modalities (graphemes and phonemes here) when a massive amount of unpaired data is available, paving the path to adopting this principled approach for all seq2seq models in low data resource regimes.

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。

↑