Variational Message Passing with Structured Inference Networks
Variational Message Passing with Structured Inference Networks
复制标题
结构化推理网络的变分消息传递
DOI：
--
复制
发表时间：
2018
期刊：
影响因子：
--
通讯作者：
M. E. Khan
中科院分区：
文献类型：
--
作者：
Wu Lin;Nicolas Hubacher;M. E. Khan
最近将深度模型与概率图模型相结合的努力在提供灵活的模型方面很有希望，这些模型也很容易解释。我们提出了一个变分的消息传递算法变分推理在这样的模型。我们做了三个贡献。首先，我们提出了结构化的推理网络，它将图形模型的结构纳入变分自动编码器（VAE）的推理网络中。其次，我们建立条件下，这样的推理网络，使快速摊销推理类似于VAE。最后，我们推导出一个变分的消息传递算法进行有效的自然梯度推理，同时保持摊销推理的效率。通过同时启用深度结构化模型的结构化，摊销和自然梯度推理，我们的方法简化和推广了现有的方法。
Recent efforts on combining deep models with probabilistic graphical models are promising in providing flexible models that are also easy to interpret. We propose a variational message-passing algorithm for variational inference in such models. We make three contributions. First, we propose structured inference networks that incorporate the structure of the graphical model in the inference network of variational auto-encoders (VAE). Second, we establish conditions under which such inference networks enable fast amortized inference similar to VAE. Finally, we derive a variational message passing algorithm to perform efficient natural-gradient inference while retaining the efficiency of the amortized inference. By simultaneously enabling structured, amortized, and natural-gradient inference for deep structured models, our method simplifies and generalizes existing methods.