针对普遍边缘化者的掩蔽计划
Masking schemes for universal marginalisers
摘要：
我们考虑结构不可知和结构相关的掩蔽方案的效果时，训练一个通用的边缘化器（arXiv：1711.00695），以学习条件分布的形式$P(x_i |\mathbf x_{\mathbf b})$，其中$x_i$是一个给定的随机变量和$\mathbf x_{\mathbf b}$是一些任意子集的所有随机变量的生成模型的兴趣。换句话说，我们模仿去噪自动编码器的自监督训练，其中未标记数据的数据集被用作部分观察输入，神经近似器被优化以最大限度地减少重建损失。我们专注于研究部分观测数据的基本过程-当预测时的观测过程与训练过程中的掩蔽过程不同时，神经近似器在学习所有条件分布方面有多好？我们比较了用不同掩蔽方案训练的网络的预测性能和泛化性能。
英文摘要：
We consider the effect of structure-agnostic and structure-dependent masking schemes when training a universal marginaliser (arXiv:1711.00695) in order to learn conditional distributions of the form $P(x_i |\mathbf x_{\mathbf b})$, where $x_i$ is a given random variable and $\mathbf x_{\mathbf b}$ is some arbitrary subset of all random variables of the generative model of interest. In other words, we mimic the self-supervised training of a denoising autoencoder, where a dataset of unlabelled data is used as partially observed input and the neural approximator is optimised to minimise reconstruction loss. We focus on studying the underlying process of the partially observed data---how good is the neural approximator at learning all conditional distributions when the observation process at prediction time differs from the masking process during training? We compare networks trained with different masking schemes in terms of their predictive performance and generalisation properties.
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。
