Krisztián Balázs

Comparison of Various Evolutionary and Memetic Algorithms

Optimization methods known from the literature include gradient based techniques and evolutionary algorithms. The main idea of the former methods is to calculate the gradient of the objective function at the actual point and then to step towards better values according to this function value. Evolutionary algorithms imitate a simplified abstract model of evolution observed in the nature. Memetic algorithms traditionally combine evolutionary and other, e.g. gradient techniques to exploit the advantages of both methods. Our current research aims to discover the properties, especially the efficiency (i.e. the speed of convergence) of particular evolutionary and memetic algorithms. For this purpose the techniques are compared by applying them on several numerical optimization benchmark functions and on machine learning problems.

Comparative Investigation of Various Evolutionary and Memetic Algorithms

Optimization methods known from the literature include gradient techniques and evolutionary algorithms. The main idea of gradient methods is to calculate the gradient of the objective function at the actual point and then to step towards better values according to this value. Evolutionary algorithms imitate a simplified abstract model of evolution observed in nature. Memetic algorithms traditionally combine evolutionary and gradient techniques to exploit the advantages of both methods. Our current research aims to discover the properties, especially the efficiency (i.e. the speed of convergence) of particular evolutionary and memetic algorithms. For this purpose the techniques are compared on several numerical optimization benchmark functions and on machine learning problems.

Comparative analysis of interpolative and non-interpolative fuzzy rule based machine learning systems applying various numerical optimization methods

In this paper interpolative and non-interpolative fuzzy rule based machine learning systems are investigated by using simulation results. The investigation focuses mainly on two objectives: to compare the efficiency of the inference techniques combined with different numerical optimization methods for solving machine learning problems and to discover the difference between the properties of systems applying interpolative and non-interpolative inference techniques.

Hybrid Bacterial Iterated Greedy heuristics for the Permutation Flow Shop Problem

This paper proposes approaches for combining the Iterated Greedy (IG) technique, as a presently state-of-the-art method, with a recently proposed adapted version of the Bacterial Evolutionary Algorithm (BEA) in order to efficiently solve the Permutation Flow Shop Problem. The obtained techniques are evaluated via simulation runs carried out on the well-known Taillards benchmark problem set. Based on the experimental results the hybrid methods are compared to each other and to the original techniques (i.e. to the original IG and BEA algorithms).

Hierarchical-interpolative fuzzy system construction by Genetic and Bacterial Programming Algorithms

In this paper a method is proposed for constructing hierarchical-interpolative fuzzy rule bases in order to model black box systems defined by input-output pairs, i.e. to solve supervised machine learning problems. The resulting hierarchical rule base is the knowledge base, which is constructed by using evolutionary techniques, namely, Genetic and Bacterial Programming Algorithms. Applying hierarchical-interpolative fuzzy rule bases is an advanced way of reducing the complexity of the knowledge base, whereas evolutionary methods ensure a relatively efficient learning process. This is the reason of the investigation of this combination.

Hierarchical fuzzy system modeling by Genetic and Bacterial Programming approaches

In this paper a method is proposed for constructing hierarchical fuzzy rule bases in order to model black box systems defined by input-output pairs, i.e. to solve supervised machine learning problems. The resultant hierarchical rule base is the knowledge base, which is constructed by using structure constructing evolutionary techniques, namely, Genetic and Bacterial Programming Algorithms. Applying hierarchical fuzzy rule bases is a way of reducing the complexity of the knowledge base, whereas evolutionary methods ensure a relatively efficient learning process. This is the reason of the investigation of this combination.

Hierarchical-interpolative fuzzy system construction by genetic and bacterial memetic programming approaches

In this paper a family of new methods are proposed for constructing hierarchical-interpolative fuzzy rule bases in the frame of a fuzzy rule based supervised machine learning system modeling black box systems defined by input-output pairs. The resulting hierarchical rule base is constructed by using structure building pure evolutionary and memetic techniques, namely, Genetic and Bacterial Programming Algorithms and their memetic variants containing local search steps. Applying hierarchical-interpolative fuzzy rule bases is a rather efficient way of reducing the complexity of knowledge bases, whereas evolutionary methods (including memetic techniques) ensure a relatively fast convergence in the learning process. As it is presented in the paper, by applying a newly proposed representation schema these approaches can be combined to form hierarchical-interpolative machine learning systems.

A Remark on Adaptive Scheduling of Optimization Algorithms

In this paper the scheduling problem of optimization algorithms is defined. This problem is about scheduling numerical optimization methods from a set of iterative ’oracle-based’ techniques in order to obtain an as efficient as possible optimization process based on the given set of algorithms.

Genetic and Bacterial Memetic Programming approaches in hierarchical-interpolative fuzzy system construction

As a straightforward continuation of our previous work in this paper new memetic (combined evolutionary and gradient based) methods are proposed for constructing hierarchical-interpolative fuzzy rule bases in the frame of a supervised machine learning system modeling black box systems defined by input-output pairs. In this work the resulting hierarchical rule bases are constructed by using structure building Genetic and Bacterial Memetic Programming Algorithms, thus stochastic evolutionary optimization methods containing deterministic local search steps. Applying hierarchical-interpolative fuzzy rule bases has proved an efficient way of reducing the complexity of knowledge bases, whereas memetic techniques often ensure a relatively fast convergence in the learning process. The literature has highlighted the advantages of memetic methods against pure evolutionary algorithms, thus the combination of hierarchical-interpolative fuzzy rule bases with memetic techniques may form promising hierarchical-interpolative machine learning systems.

Adaptive scheduling of optimization algorithms in the construction of interpolative fuzzy systems

This paper presents an adaptive scheduling approach applied for constructing interpolative fuzzy rule based systems. This is a continuation of our preceding work, where the same approach was used for dense fuzzy rule bases. During the optimization process different optimization algorithms are scheduled according to their respective local efficiency, i.e. according to their convergence speed in various phases of the machine learning process. The scheduled optimization techniques are evolutionary algorithms that have shown efficiency in the construction of interpolative fuzzy rule based systems. Simulations are carried out on standard benchmark sets in order to evaluate the established system and to compare it to fuzzy systems built up by deploying the same optimization techniques without the scheduling approach.