An Adaptive Resource Allocating Neuro-Fuzzy Inference System with Sensitivity Analysis Resource Control

Abstract

Adaptability in non-stationary contexts is a very important property and a constant desire for modern intelligent systems and is usually associated with dynamic system behaviors. In this framework, we present a novel methodology of dynamic resource control and optimization for neurofuzzy inference systems. Our approach involves a neurofuzzy model with structural learning capabilities that adds rule nodes when necessary during the training phase. Sensitivity analysis is then applied to the trained network so as to evaluate the network rules and control their usage in a dynamic manner based on a confidence threshold. Therefore, on one hand, we result in a well-balanced structure with an improved adaptive behavior and, on the other hand, we propose a way to control and restrict the "curse of dimensionality". The experimental results on a number of classification problems prove clearly the strengths and benefits of this approach.

Simulations and fuzzy systems for resource reallocation

“…In Goudarzi et al, 1 a fuzzy controller is used to detect the bottlenecks that occur in telecommunications services and to try to improve speed, since the fuzzy system proposed allows for resources to be redistributed and does not require feedback, relying instead on previous and current estimates. Pertselakis et al 2 use neurofuzzy systems to redistribute resources via a neurofuzzy architecture that combines resource distribution procedures and fuzzy sets to incorporate inference and expert knowledge. In addition, by being general, it can be used in fields where there are no guidelines on the rules needed to resolve particular problems by using a dynamic architecture to provide the methods for modeling non-stationary phenomena.…”

A New Parameter Determining Mechanism for Radial Basis Neural Networks

“…2) Comparing results with past researches carried out to initialize the parameters of radial basis neural networks: The table II presents some past results and the results of Paul and Kumar (2002) [15] and Pertselakis [11].These results are extracted from the paper by Manolis Wallacea, Nicolas Tsapatsoulisb and Stefanos Kollias. [17].…”

Υβριδικές Ευφυείς Τεχνικές Ανάλυσης Για Εξόρυξη Γνώσης Από Δεδομένα Υψηλών Διαστάσεων

“…Η αρχιτεκτονική και η λειτουργικότητα του νευρο-ασαφούς μοντέλου βασί ζεται στο μοντέλο ARANFIS [61], το οποίο εφαρμόζει την λογική RAN [62] στο μοντέλο SuPFuNIS [60]. Η αρχιτεκτονική περιλαμβάνει τρία επίπεδα: το επίπεδο εισόδου, το επίπεδο των κανόνων και το επίπεδο εξόδου.…”

“…όπου η(ί) είναι ο προσαρμοζόμενος ρυθμός μάθησης και είναι μία παράμετρος που εισάγεται από την παρούσα μεθοδολογία. Οι αναλυτικές εξισώσεις των μερικών παραγώγων περιγράφονται στο [61].…”