LOSS FUNCTIONS AND DESCENT METHOD

Victor K. Ohanyan; Hovhannes Z. Zohrabyan

doi:10.46991/PYSU:A/2021.55.1.029

LOSS FUNCTIONS AND DESCENT METHOD

Authors

Victor K. Ohanyan Chair of the Theory of Probability and Mathematical Statistics, YSU, Armenia
Hovhannes Z. Zohrabyan Chair of the Theory of Probability and Mathematical Statistics, YSU, Armenia

DOI:

https://doi.org/10.46991/PYSU:A/2021.55.1.029

Keywords:

Bayesian estimators, gradient descent, loss functions, machine learning

Abstract

In this paper, we showed that it is possible to use gradient descent method to get minimal error values of loss functions close to their Bayesian estimators. We calculated Bayesian estimators mathematically for different loss functions and tested them using gradient descent algorithm. This algorithm, working on Normal and Poisson distributions showed that it is possible to find minimal error values without having Bayesian estimators. Using Python, we tested the theory on loss functions with known Bayesian estimators as well as another loss functions, getting results proving the theory.

Downloads

Published

2021-05-21

How to Cite

Ohanyan, V. K., & Zohrabyan, H. Z. (2021). LOSS FUNCTIONS AND DESCENT METHOD. Proceedings of the YSU A: Physical and Mathematical Sciences, 55(1 (254), 29–35. https://doi.org/10.46991/PYSU:A/2021.55.1.029