Resilience Pattern in Neural Network

Please enable javascript in your browser.

Mozafari, Tahereh | 2017

802 Viewed

Type of Document: M.Sc. Thesis
Language: Farsi
Document No: 50821 (04)
University: Sharif University of Technology
Department: Physics
Advisor(s): Rahimi Tabar, Mohammad Reza
Abstract:
In this thesis the resilience of a neural network is examined under structural perturbation.The network consists of Spiking neurons with specific dynamics or they are binary neurons and the network is non-spiking. Neural network is a multi-dimensional and multi-stable system therefore we can map their dynamical equations to effective one-dimensional equation with one parameter which shows the environmental conditions.The resilience of the system is evaluated with respect to changes in weights and edges of the network and the the resilience pattern of the system is obtained.Finally, the recovery and control methods are briefly discussed
Keywords:
Resilience ; Spiking Neural Network ; Artificial Neural Network ; Recovery ; Control