SYNCHRONIZATION BETWEEN INTEGER AND FRACTIONAL ORDER DELAY NEURAL NETWORKS SYSTEM

Fatin Nabila Abd Latiff,Wan Ainun Mior Othman Wan Ainun Mior Othman

doi:10.31580/sps.v2i2.1330

Fatin Nabila Abd Latiff, Wan Ainun Mior Othman Wan Ainun Mior Othman

Open Access

https://doi.org/10.31580/sps.v2i2.1330

Copy DOI

Journal: Science proceedings series	Publication Date: May 18, 2020
License type: CC BY 4.0

Affiliation: University of Malaya

Abstract

The purpose is to present a method for synchronizing a recurrent neural networks system between integer and fractional-order order delay by active sliding mode control . The Active Sliding Mode Control (ASMC) scheme is used to solve the synchronization problem between the integer-order delayed recurrent neural networks system via active sliding mode control (IoDRNNASM) systems and the fractional-order delay recurrent neural networks system via active sliding mode control (FoDRNNASM) system based on the Lyapunov direct fractional method (LDFM). To explore the behavior of the IoDRNNASM systems and the FoDRNNASM systems, we performed the technique of numerical simulations using MATLAB software to prove the feasibility and strength of the archived outcomes. This concept can also be enhanced with the implementation of double encryption using RSA encryption to secure communication. Because we expected in the future that this enhanced concept will strengthen and increase the network security capabilities that will provide powerful protection in secure communications.

Highlights

To overcome the activation function problem, a class of novel Lyapunov fractional functions is established to indicate that the synchronization error asymptotically converges to zero
To proposed that the neural networks of fractional order and integer order are synchronized globally with time delays relying on a novel Lyapunov fractional function and delayed active sliding mode controller (DASMC)
IoDRNNASMC acts as a master system while FoDRNNASMC acts as a slave system

Summary

Introduction

To overcome the activation function problem, a class of novel Lyapunov fractional functions is established to indicate that the synchronization error asymptotically converges to zero. The systems are synchronized using ASMC and LDFM. Systems are synchronised in the form of delayed systems. The implementation of the master-slave system in the secure communication system.

Results

Conclusion