Activation of Frictional Functions In Neural Artificial Networks

  • G.R.Mahendra Babu ,Dr.S.Gopinath ,Mr.E.Arunkumar

Abstract

Fractional calculus, including physics, biology and artificial intelligence, is an important tool for analysis. Alternative fractional definitions of the exponential function, trigonometric functions, hyperbolic tangent have been developed for some functions. Yet fractional parallels are not yet studied for others, such as logistic sigmoid function. Most of the above functions are used in Artificial Neural Networks As Activation Functions (ANNAF). The ANNAF networks have more tunable hyperparameters through the use of fractional activation functions. In this article, the effect on the learning and precision of feed artificial neural networks is investigated and the results of different choices of function parameter values are studied.

Published
2020-04-18
How to Cite
G.R.Mahendra Babu ,Dr.S.Gopinath ,Mr.E.Arunkumar. (2020). Activation of Frictional Functions In Neural Artificial Networks. International Journal of Advanced Science and Technology, 29(8s), 500 - 506. Retrieved from http://sersc.org/journals/index.php/IJAST/article/view/10539