A Controlled Perturbation Algorithm for Saddle Point Escape of Generic Non-convex Optimization Problems: Deterministic Optimization Problems (Algorithms and Experiments)
- Publié
- Serveur de preprints
- Zenodo
- DOI
- 10.5281/zenodo.20083396
This preprint is the experimental results of deterministic optimization problems in the preprint "Controlled Perturbation Algorithms for Saddle Point Escape of Generic Non-convex Optimization Problems (Algorithm Description – Version 1.1)" in
https://doi.org/10.5281/zenodo.19948634
5 Experiments, each with two methods, are performed to test the ability of the Controlled Perturbation Algorithm (CPA) to
escape saddle point when gradient is very small (Section 5)
work as a stanalone optimizer (Section 6)
Probabilitistic optimization problems (like machine learning and deep learning) are under progress.
The source code is provided with this preprint. If any errors are discovered, the author would appreciate being notified by email at khcheng920911@gmail.com .