Spectral-Proximal-Method

Pytorch implementation of Spectral Proximal (SP) method with Saliency Matrix

SP method is a Hybrid Proximal Method with Spectral Gradient Techniques. The method have similar theoretical convergence as Multiple Damping Gradient (MDG) method.

Citation

@article{Yong2023,
  doi = {10.22541/au.167350875.57067000/v1},
  url = {https://doi.org/10.22541/au.167350875.57067000/v1},
  year = {2023},
  month = jan,
  publisher = {Authorea,  Inc.},
  author = {Cherng-Liin Yong and Ban-Hoe Kwan and Danny-Wee-Kiat Ng and Hong Seng Sim},
  title = {Robust Optimization of Deep Learning Models using Spectral Proximal Method and Saliency Matrix}
}

@INPROCEEDINGS{9377294,
  author={Yong, Cherng Liin and Kwan, Ban Hoe and Ng, Danny Wee-Kiat and Sim, Hong Seng},
  booktitle={2021 IEEE 17th International Colloquium on Signal Processing & Its Applications (CSPA)}, 
  title={Optimized Machine Learning Algorithm using Hybrid Proximal Method with Spectral Gradient Techniques}, 
  year={2021},
  volume={},
  number={},
  pages={101-106},
  doi={10.1109/CSPA52141.2021.9377294}
}