Slide 39
Slide 39 text
Bayesian logistic regression
We perform the standard Bayesian logistic regression experiment on
the Covertype dataset, following the same settings as Liu et al.1. We
compare our methods with MCMC, SVGD, WNAG2 and WNes3.
We select the bandwidth using either the MED method or the
proposed BM method. The BM method is not applied to SVGD.
1Qiang Liu and Dilin Wang. Stein variational gradient descent: A general purpose
bayesian inference algorithm. In Advances in neural information processing systems,
pp. 23782386, 2016.
2Chang Liu, Jingwei Zhuo, Pengyu Cheng, Ruiyi Zhang, Jun Zhu, and Lawrence
Carin. Accelerated first-order methods on the wasserstein space for bayesian inference.
arXiv preprint arXiv:1807.01750, 2018.
3Chang Liu, Jingwei Zhuo, Pengyu Cheng, Ruiyi Zhang, and Jun Zhu.
Understanding and accelerating particle-based variational inference. In International
Conference on Machine Learning, pp. 40824092, 2019.
39