ZhuSuan: A Library for Bayesian Deep Learning

By Jiaxin Shi, Jianfei Chen, Jun Zhu, Yucen Luo, Yihong Gu, Yuhao Zhou. 2017

43 citations recorded by the source corpus.

In this paper we introduce ZhuSuan, a python probabilistic programming library for Bayesian deep learning, which conjoins the complimentary advantages of Bayesian methods and deep learning. ZhuSuan is built upon Tensorflow. Unlike existing deep learning libraries, which are mainly designed for deterministic neural networks and supervised tasks, ZhuSuan is featured for its deep root into Bayesian inference, thus supporting various kinds of probabilistic models, including both the traditional hierarchical Bayesian models and recent deep generative models. We use running examples to illustrate the probabilistic programming on ZhuSuan, including Bayesian logistic regression, variational auto-encoders, deep sigmoid belief networks and Bayesian recurrent neural networks.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Zhiyuan Liu
2. Minlie Huang
3. Jun Zhu
4. Xipeng Qiu
5. Jiaxin Shi
6. Maosong Sun
7. Yingzhen Li
8. Mohit Bansal

In Computer Vision and Pattern Recognition

1. Jun Zhu
2. Jiaxin Shi
3. Luc Van Gool
4. Dacheng Tao
5. Yingzhen Li
6. Trevor Darrell
7. Dustin Tran
8. Xiaogang Wang

Citations in the corpus

• Pre-Trained Models: Past, Present and Future (2021, background): “Recently, much work has been done on the theory, algorithms and programming libraries of Bayesian deep learning, which conjoins Bayesian methods and deep networks (e.g., see (Shi et al., 2017) for more details).”
• Compiling Stan to generative probabilistic languages and extension to deep probabilistic programming (2021, background): “In recent years, taking advantage of the maturity of DL frameworks, multiple deep probabilistic programming languages have been proposed: PyMC3 [28] built on top of Theano, Edward [33] and ZhuSuan [29] built on top of TensorFlow, and Pyro [3] and ProbTorch [22] built on top of PyTorch.”
• A Spectral Approach to Gradient Estimation for Implicit Distributions (2018, unknown): “Implementations are based on ZhuSuan (Shi et al., 2017).”
• A review of machine learning in processing remote sensing data for mineral exploration (2021, background): “Recently, there has been tremendous progress in the area of Bayesian neural networks and Bayesian deep learning (Wang and Yeung, 2020; Shi et al., 2017); however, these methods have not been much used for remote sensing, and their application for mineral exploration is absent.”
• Variational Implicit Processes (2018, methodology): “• Variational Gaussian inference for BNN (VI-BNN): we implement VI for BNN using the Bayesian deep learning library, ZhuSuan [82].”
• TensorFlow Distributions (2017, methodology): “Further, Distribution and Bijector is being used as the design basis for similar functionality in the PyTorch computational graph framework [39], as well as the Pyro and ZhuSuan probabilistic programming systems [38, 43].”
• Simple, Distributed, and Accelerated Probabilistic Programming (2018, methodology): “To specify random choices in the program, we use RandomVariables from Edward [49], which has similarly been built on by Zhusuan [42] and Probtorch [34].”
• Kernel Implicit Variational Inference (2017, methodology): “Except a few simple examples, Bayesian inference is typically challenging, for which variational inference (VI) has been a standard workhorse to approximate the true posterior (Zhu et al., 2017).”