Python implementation of the Structured Deep Kernel Networks (SDKN).
## Usage:
The SDKN is implemented based on pytorch.
- The file main_sdkn_vs_nn.py compares the SDKN with NN on some toy example.
- The files main_lightning_NN.py and main_lightning_SDKN.py provide implementations of a NN and SDKN using pytorch-lightning.
In order to set up the venv, run ./setup.sh (maybe run 'chmod +x setup.sh' before).
## How to cite:
If you use this code in your work, please cite the paper
> T. Wenzel and G. Santin and B. Haasdonk, [_Universality and {O}ptimality of {S}tructured {D}eep {K}ernel {N}etworks_](https://arxiv.org/abs/2103.13655), ArXiv preprint 2105.07228 (2021).
```bibtex:
@article{wenzel2021universality,
title={Universality and {O}ptimality of {S}tructured {D}eep {K}ernel {N}etworks},
author={Tizian Wenzel and Gabriele Santin and Bernard Haasdonk},
journal={ArXiv},
year={2021},
type = {ArXiv},
number = {2105.07228},
url = {https://arxiv.org/abs/2105.07228},
note={Submitted}
}
```
## How to view the tensorboard logs:
In order to view tb_logs:
tensorboard --logdir=path_to_dir # view one folder
