Santa
Code for the Santa algorithm for deep learning
The Code incldues a MATLAB version for the FNN and CNN, a Python version for the RNN, and an implementation in Caffe for large-scale learning. The algorithm is proposed in the AISTATS 2016 paper "Bridging the Gap between Stochastic Gradient MCMC and Stochastic Optimization".
Changyou Chen (cchangyou@gmail.com), 2.24.2016
License
Please note that this code should be used at your own risk. There is no implied guarantee that it will not do anything stupid. Permission is granted to use and modify the code.
Usage
Please refer to README.pdf for instructions on how to use the code.
Citing Santa
Please cite our AISTATS paper in your publications if it helps your research:
@inproceedings{Santa_AISTATS2016,
Author = {C. Chen, D. Carlson, Z. Gan, C. Li, and L. Carin},
Title = {Bridging the Gap between Stochastic Gradient MCMC and Stochastic Optimization},
booktitle={AISTATS},
Year = {2016}
}