Semi-Implicit Stochastic Recurrent Neural Networks
Semi-Implicit Stochastic Recurrent Neural Networks
复制标题
DOI：
10.1109/icassp40776.2020.9053491
复制
发表时间：
2019-10
期刊：
影响因子：
--
通讯作者：
Ehsan Hajiramezanali;Arman Hasanzadeh;N. Duffield;K. Narayanan;Mingyuan Zhou;Xiaoning Qian
中科院分区：
文献类型：
--
作者：
Ehsan Hajiramezanali;Arman Hasanzadeh;N. Duffield;K. Narayanan;Mingyuan Zhou;Xiaoning Qian
具有复杂依赖结构的潜在随机变量的随机递归神经网络在建模序列数据方面比确定性深度模型更成功。然而，大多数现有的方法有有限的表达能力，由于高斯假设的潜变量。在本文中，我们提倡使用半隐式变分推理来学习隐式潜在表示，以进一步增加模型的灵活性。半隐式随机递归神经网络（SIS-RNN）的发展，以丰富推断模型后验，可能没有解析密度函数，只要独立的随机样本可以通过重新参数化生成。在真实世界数据集上进行的不同任务的大量实验表明，SIS-RNN优于现有方法。
Stochastic recurrent neural networks with latent random variables of complex dependency structures have shown to be more successful in modeling sequential data than deterministic deep models. However, the majority of existing methods have limited expressive power due to the Gaussian assumption of latent variables. In this paper, we advocate learning implicit latent representations using semi-implicit variational inference to further increase model flexibility. Semi-implicit stochastic recurrent neural network (SIS-RNN) is developed to enrich inferred model posteriors that may have no analytic density functions, as long as independent random samples can be generated via reparameterization. Extensive experiments in different tasks on real-world datasets show that SIS-RNN outperforms the existing methods.