This folder contains the Python implementation of the VBNN model presented in the submitted manuscript "Critical Temperature Prediction for a Superconductor: A Variational Bayesian Neural Network Approach".
===================================================
I. Installation:
This code depends on ZhuSuan package[1], variational Bayesian inference [2], Stochastic GradientVariational Bayes (SGVB) [3].
Note: We used Tensorflow version 1.14 with several new files and functions added to support the new model. The following packages are required additionaly to run the code. For details on installing Tensorflow, and neccessary package , please refer to:
TensorFlow: https://www.tensorflow.org/install
ZhuSuan Bayesian Deep Learning: https://zhusuan.readthedocs.io/en/latest/
II. Superconducting Material Database (SuperCon) from this page: https://supercon.nims.go.jp/index_en.html
===================================================
III. Test Result
A test result script is given in the file 'VBNN_HTc.ipynb'. It can be viewed by the link: https://github.com/ltdung/VBNN_HighTc/blob/master/VBNN_HTc.ipynb. In cases of, you find difficulty to view the file, please use: https://nbviewer.jupyter.org/
=====================================================
References
[1] J. Shi, J. Chen, J. Zhu, S. Sun, Y. Luo, Y. Gu, and Y. Zhou, “ZhuSuan:A library for Bayesian deep learning,” arXiv preprint arXiv:1709.05870, 2017.
[2] J. Drugowitsch, “Variational bayesian inference for linear and logistic regression,” arXiv preprint arXiv:1310.5438, 2013
[3] D. P. Kingma and M. Welling, “Auto-encoding variational bayes,” arXivpreprint arXiv:1312.6114, 2013