Fitness Function Design Research Articles

A multi-objective evolutionary approach to automatic melody generation

Abstract Existing evolutionary approaches to automatic composition generate only a few melodies in a certain style that is specified by the setting of parameters or the design of fitness functions. Thus, their composition results cannot cover the various tastes of music. In addition, they are not able to deal with the multidimensional nature of music. This paper presents a novel multi-objective evolutionary approach to automatic melody composition in order to produce a variety of melodies at once. To this end, two conflicting fitness measures are investigated to evaluate the fitness of melody; (1) stability and (2) tension . Resorting to music theory, genetic operators (i.e., crossover and mutation) are newly designed to improve search capability in the multi-objective fitness space of music composition. The experimental results demonstrate the validity and effectiveness of the proposed approach. Moreover, the analysis of composition results proves that the proposed approach generates a set of pleasant and diverse melodies.

Handling swarm of UAVs based on evolutionary multi-objective optimization

The fast technological improvements in unmanned aerial vehicles (UAVs) has created new scenarios where a swarm of UAVs could operate in a distributed way. This swarm of vehicles needs to be controlled from a set of ground control stations, and new reliable mission planning systems, which should be able to handle the large amount of variables and constraints. This paper presents a new approach where this complex problem has been modelled as a constraint satisfaction problem (CSP), and is solved using a multi-objective genetic algorithm (MOGA). The algorithm has been designed to minimize several variables of the mission, such as the fuel consumption or the makespan among others. The designed fitness function, used by the algorithm, takes into consideration, as a weighted penalty function, the number of constraints fulfilled for each solution. Therefore, the MOGA algorithm is able to manage the number of constraints fulfilled by the selected plan, so it is possible to maximize in the elitism phase of the MOGA the quality of the solutions found. This approach allows to alleviate the computational effort carried out by the CSP solver, finding new solutions from the Pareto front, and therefore reducing the execution time to obtain a solution. In order to test the performance of this new approach 16 different mission scenarios have been designed. The experimental results show that the approach outperforms the convergence of the algorithm in terms of number of generations and runtime.

Hierarchic Entropy: An Information Theoretic Measure of Evolutionary Robotic Behavioral Diversity

This paper investigates the design of information theoretic-based fitness function for embedded evolutionary robotics (ERs). Such fitness relies on the assumption that interesting behaviors result in a high sensorimotor (individual) diversity. The current simple entropy as a diversity metric only considers individuals’ difference but ignores their spatial relationship. The sensorimotor stream can be analyzed to construct a simple directed graph that has unique entry and exit nodes. This paper proposes a hierarchic entropy as a diversity metric by incorporating the simple entropy and the spatial relationship based graph entropy. Maximizing the hierarchic entropy, achieved by on-board evolutionary algorithm, thus defines a self-driven fitness function enforcing the controller visiting diverse sensorimotor states. The proposed algorithm achieves better performance than the published results of other entropy-based methods only relying on simple entropy, without requiring additional computational resources.

Optimised placement of wireless sensor networks by evolutionary algorithm

In wireless sensor networks, computing the sensor nodes collection that meets the requirements is called the problem of network coverage optimisation for wireless sensor network; also called the problem of minimum connected covering node set. The innovation points of this article are as follows: Firstly, it analyses the deficiencies of traditional evolution algorithm fitness function, puts forward an improved fitness function design scheme, and proves that it has advantage of solving problem on wireless sensor networks coverage optimisation. Secondly, it applies the method of control variables, comparison and analysis of the influence on the various operations and parameters selection in evolution algorithms on the optimisation results and performance, and then points out how to design algorithms to best optimise effect and performance.

Differential evolution based high-order peak filter design with application to compensation of contact-induced vibration in HDD servo systems

Sensitivity loop shaping using add-on peak filters is a simple and effective method to reject narrow-band disturbances in hard disk drive (HDD) servo systems. The parallel peak filter is introduced to provide high-gain magnitude in the concerned frequency range of open-loop transfer function. Different from almost all the known peak filters that possess second-order structures, we explore in this paper how high-order peak filters can be designed to improve the loop shaping performance. The main idea is to replace some of the constant coefficients of common second-order peak filter by frequency-related transfer functions, and then differential evolution (DE) algorithm is adopted to perform optimal design. We creatively introduce chromosome coding and fitness function design, which are original and the key steps that lead to the success of DE applications in control system design. In other words, DE is modified to achieve a novel design for hard disk drive control. Owing to the remarkable searching ability of DE, the expected shape of sensitivity function can be achieved by incorporating the resultant high-order peak filter in parallel with baseline feedback controller. As a result, a seventh-order peak filter is designed to compensate for contact-induced vibration in a high-density HDD servo system, where the benefits of high-order filter are clearly demonstrated.

Evolution or revolution: the critical need in genetic algorithm based testing

Software testing is one of the most inevitable processes in software development. The field of software testing has seen an extensive use of search based techniques in the last decade. Among the search based techniques, it is the metaheuristic techniques such as genetic algorithm that has garnered the major share of attention from researchers. Looking at the large body of work that has happened and is happening in this field, we feel that it is high time someone studied how well genetic algorithm based techniques fare in practical testing process. Method: In this work, we present a roadmap to the future of genetic algorithm based software testing, based on a review of literature. We have mainly reviewed the works which use genetic algorithm for software test data generation. This independent review is designed to direct the attention of future researchers to the deficiencies of genetic algorithm based testing, their possible solutions and the extent to which they are correctable. The observations from the selected primary studies highlight the issues faced when genetic algorithm is applied in software testing. The observations form the review reveal that the type of genetic algorithm used, fitness function design, population initialization and parameter settings does impact the quality of solution obtained in software testing using genetic algorithm. From the review we conclude that, more generalized approaches can make genetic algorithm based software testing one of the strongest methods in practical software testing. We hope that, this review will be a major breakthrough in genetic algorithm based software testing field.

Hybrid Algorithm for Clustering Gene Expression Data

Microarray gene expressions provide an insight into genomic biomarkers that aid in identifying cancerous cells and normal cells. In this study, functionally related genes are identified by partitioning gene data. Clustering is an unsupervised learning technique that partition gene data into groups based on the similarity between their expression profiles. This identifies functionally related genes. In this study, a hybrid framework is designed that uses adaptive pillar clustering algorithm and genetic algorithm. A first step towards, the proposed work is the utilization of clustering technique by adaptive pillar clustering algorithm that finds cluster centroids. The centroids and its clustering elements are calculated by average mean of pairwise inner distance. The output of adaptive pillar clustering algorithm results in number of clusters which is given as input to genetic algorithm. The microarray gene expression data set considered as input is given to adaptive pillar clustering algorithm that partitions gene data into given number of clusters so that the intra-cluster similarity is maximized and inter cluster similarity is minimized. Then for each combination of clustered gene expression, the optimum cluster is found out using genetic algorithm. The genetic algorithm initializes the population with set of clusters obtained from adaptive pillar clustering algorithm. Best chromosomes with maximum fitness are selected from the selection pool to perform genetic operations like crossover and mutation. The genetic algorithm is used to search optimum clusters based on its designed fitness function. The fitness function designed minimizes the intra cluster distance and maximizes the fitness value by tailoring a parameter that includes the weightage for diseased genes. The performance of adaptive pillar algorithm was compared with existing techniques such as k-means and pillar k-means algorithm. The clusters obtained from adaptive pillar clustering algorithm achieve a maximum cluster gain of 894.84, 812.4 and 756 for leukemia, lung and thyroid gene expression data, respectively. Further, the optimal cluster obtained by hybrid framework achieves cluster accuracy of 81.3, 80.2 and 78.2 for leukemia, lung and thyroid gene expression data respectively.

Classification of Remotely Sensed Images Using the GeneSIS Fuzzy Segmentation Algorithm

In this paper, we propose an integrated framework of the recently proposed Genetic Sequential Image Segmentation (GeneSIS) algorithm. GeneSIS segments the image in an iterative manner, whereby at each iteration, a single object is extracted via a genetic algorithm-based object extraction method. This module evaluates the fuzzy content of candidate regions, and through an effective fitness function design provides objects with optimal balance between fuzzy coverage, consistency and smoothness. GeneSIS exhibits a number of interesting properties, such as reduced over-/undersegmentation, adaptive search scale, and region-based search. To enhance the capabilities of GeneSIS, we incorporate here several improvements of our initial proposal. On one hand, two modifications are introduced pertaining to the object extraction algorithm. Specifically, we consider a more flexible representation of the structural elements used for the object's extraction. Furthermore, in view of its importance, the consistency criterion is redefined, thus providing a better handling of the ambiguous areas of the image. On the other hand we incorporate three tools properly devised, according to the fuzzy principles characterizing GeneSIS. First, we develop a marker selection strategy that creates reliable markers, particularly when dealing with ambiguous components of the image. Furthermore, using GeneSIS as the essential part, we consider a generalized experimental setup embracing two different classification schemes for remote sensing images: the spectral-spatial classification and the supervised segmentation methods. Finally, exploiting the inherent property of GeneSIS to produce multiple segmentations, we propose a segmentation fusion scheme. The effectiveness of the proposed methodology is validated after thorough experimentation on four data sets.

Voltage Flicker Parameters Estimation UsingShuffled Frog Leaping Algorithm (SFLA) andImperialistic Competitive Algorithm (ICA)

Measurement of magnitude and frequency ofthe voltage flicker is very important for monitoring and controlling voltage flicker efficiently to improve the network power quality. This paper presents two new methods for measurement of flicker signal parameters using Shuffled Frog Leaping Algorithm (SFLA) and Imperialist Competitive Algorithm (ICA). This paper estimates fundamental voltage and flicker magnitudes and frequencies with proposed methods. The goal is to minimize the error ofthe estimated magnitudes and frequencies via a designed fitness function. At first we introduce voltage flicker and its measuring techniques. Then voltage flicker model is analyzed. At the next part a review of SFLA and ICA is presented these methods are applied to a test voltage signal and the results are given and analyzed.

Multi-objective fractional artificial bee colony algorithm to energy aware routing protocol in wireless sensor network

Due to the promising application of collecting information from remote or inaccessible location, wireless sensor networks pose big challenge for data routing to maximize the communication with more energy efficient. Literature presents different cluster-based energy aware routing protocol for maximizing the life time of sensor nodes. Accordingly, an energy efficient clustering mechanism, based on artificial bee colony algorithm and factional calculus is proposed in this paper to maximize the network energy and life time of nodes by optimally selecting cluster-head. The hybrid optimization algorithm called, multi-objective fractional artificial bee colony is developed to control the convergence rate of ABC with the newly designed fitness function which considered three objectives like, energy consumption, distance travelled and delays to minimize the overall objective. The performance of the proposed FABC-based cluster head selection is compared with LEACH, PSO and ABC-based routing using life time, and energy. The results proved that the proposed FABC maximizes the energy as well as life time of nodes as compared with existing protocols.

Stability Analysis of Tunnel - Slope Coupling Based on Genetic Algorithm

Subjects in tunnels, being constrained by terrain and routes, entrances and exits to tunnels, usually stay in the terrain with slopes. Thus, it is necessary to carry out stability analysis by treating the tunnel slope as an entity. In this study, based on the Janbu slices method, a model for the calculation of the stability of the original slope-tunnel-bank slope is established. The genetic algorithm is used to implement calculation variables, safety coefficient expression and fitness function design. The stability of the original slope-tunnel-bank slope under different conditions is calculated, after utilizing the secondary development function of the mathematical tool MATLAB for programming. We found that the bearing capacity of the original slopes is reduced as the tunnels are excavated and the safety coefficient is gradually decreased as loads of the embankment construction increased. After the embankment was constructed, the safety coefficient was 1.38, which is larger than the 1.3 value specified by China’s standards. Thus, the original slope-tunnel-bank slope would remain in a stable state.

A Multiple Objective Fitness Function Design Scheme for GA-Based Cognitive Engines

In this paper, we propose a multiple objective fitness function for cognitive engines by using the genetic algorithm (GA). Specifically, we propose four single objective fitness functions, and finally, we propose a multiple objective fitness function based on the single objective fitness functions for transmission parameter optimization. Numerical results demonstrate that we can obtain transmission parameter sets optimized for given transmission scenarios with the GA-based cognitive engine incorporating the proposed objective fitness function.

Test Paper Composition Design on College English Examination Based on Genetic Algorithm

College English examination is an important part of university English education, for the difficult problems in the process of examination organization and management, in this paper, the intelligent test paper composition research based on genetic algorithm. First, basic research, including the basic idea of genetic algorithm, algorithm flow and perform operations on groups; then, studies test paper composition model, the test paper problem described as question number, question types, difficulty, discrimination, score, answer time, using the frequency property and other attributes, constituted seven dimensional vector. Finally, test paper composition design, to generate the initial population, fitness function design, operation operator design, algorithm terminates and others, four key steps were designed. The content of this paper to meet the quality and speed requirements of university English test paper composition, and has the advantages of randomness, scientificalness and so on.

GeneSIS: A GA-based fuzzy segmentation algorithm for remote sensing images

This paper proposes an object-based classification scheme for handling remotely sensed images. The method combines the results of a supervised pixel-based classifier with spatial information extracted from image segmentation. First, pixel-wise classification is implemented by a fuzzy output SVM classifier using spectral and textural features of pixels. This classification results to a set of fuzzy membership maps. Operating on this transformed space, a Genetic Sequential Image Segmentation (GeneSIS) algorithm is next developed to partition the image into homogeneous regions. GeneSIS follows a sequential object extraction approach, whereby at each iteration a single object is extracted by invoking a GA-based object extraction algorithm. This module evaluates the fuzzy content of candidate regions, and through an effective fitness function design provides objects with optimal balance between three fuzzy components: coverage, consistency and smoothness. The final classification map is obtained automatically via segmentation, since each segment is extracted with its own class label. The validity of the proposed method is shown on the land cover classification of three different remote sensing images, with varying number of spectral bands (multispectral/hyperspectral), different spatial resolutions and ground truth cover types. The accuracy results of our approach are favorably compared with the ones obtained by other segmentation-based classification techniques.

Mining of Optimized Multi Relational Relation Patterns for Prediction System

Prediction is arguably considered the main goal of data mining, with the greatest potential payoff. The two principal prediction problems are classification and regression. In many data mining tools that support classification tasks, training data are stored in a single table containing both the target field and the attributes. Generally, only intra-tuple relationships between the attributes and the target field are found, while inter-tuple relationships are not considered and relationships between several tuples of distinct tables are not even able explore. Disregarding inter-table relationships can be a severe limitation in many real-world applications that involve the prediction of numerical values from data that are naturally organized in a relational model involving several tables. In this paper, we have concentrated on classification task of multi-relational data. We have planned to develop an optimized rule generation with the newly designed fitness function of genetic algorithm. Once the rule is mined, the relevant attributes will be identified from multi relational patterns. Then, the database is converted to single table with all the relevant attributes from multiple tables. Finally, the neural network will be used to training the converted data based on the association relationship. Then, in the testing phase, the prediction will be done with the help of trained neural network. In the experimental analysis, the proposed approach has gained an average accuracy of 95% in the prediction process by neural network.

Study of Network Division Optimization Based on Improved Genetic Algorithm Undirected Graph Theory

The paper analyzed the essence of network division optimization, proposed to use undirected graph division theory to study the issue and, combing the feature of the issue, designed an improved genetic algorithm. The algorithm improved the classical genetic algorithm from fitness function design, genetic operator and parameter selection. The actual research results reveal that the algorithm realized the automatic network division optimization and the performance is better than classical genetic algorithm.

Kinetic Parameters Identification of Double Pendulum Robot

The model structure of Double Pendulum Robot (DPR) has been built up with Lagrange equations. To reduce the difference between simulation control and real-time control, the kinetic parameters of Double pendulum robot (DPR) is accurately identified with proposed improved genetic algorithm (IGA) which integrated hybrid coding, hamming distance, orthogonal experiment, feedback mutation, and a specially designed fitness function. This improved genetic algorithm (IGA) overcome the shortcoming of traditional genetic algorithm, strengthen global and local searching capability, and get results more close to global optimal solution. Finally, the optimal results (kinetic parameter values) are applied to model, and make a contrast experiment of zero input response between searched optimal kinetic parameters and real kinetic parameters of DPR. DOI: http://dx.doi.org/10.11591/telkomnika.v11i3.2174

Optimization Research of Process Parameters for Laser Cladding Valve Parts Based on Genetic Algorithm

The optimization research of process parameters for big power Laser cladding valve parts, is a research focus of modern surface hardening technology. The article discussed in detail for solving the optimum process parameters of Laser cladding for the selection approach of strategy of genetic algorithm, the quantitative relationship model was established between process parameters and the valve parts property using neural network method , which process parameters are laser power (P), scanning speed (V), powder feeding rate (G), scan spacing (D) and thickness ( ) etc., the best configuration program of Genetic Algorithm control parameters has been obtain by means of the parameters encoding、initial group setting、fitness function design,genetic operation design and algorithm control parameters setting. The optimization of process parameters is obtained to fit the Laser cladding technology by using genetic algorithm toolbox in the MATLAB environment, and the optimization goal of the valve parts property has also been achieved. Practice has proved that the optimal process parameters are correct by the genetic algorithm , and has a very good production practice guide.

The Strong Prime Numbers Generation Algorithm Based on Genetic Algorithm

Large prime number generation methods in need of a more complex modular exponentiation, leading to defects of slower computing speed, Based on this genetic algorithm, proposed a new strong prime number generated algorithm. The method according to the characteristics of Strong Primes, the algorithm is simple, easy to implement to meet the needs of the RSA algorithm security , giving the method of determining the large prime numbers. Design fitness function , crossover and mutation strategies which can be used in genetic algorithm. Finally design the algorithm of producing Strong prime numbers

Parameter Optimization of PID Controller for Boiler Combustion System by Applying Adaptive Immune Genetic Algorithm

For power plant boiler combustion control system has large inertia, nonlinear and other complex characteristics, a control algorithm of PID optimized by means of adaptive immune genetic algorithm is presented. A variety of improved schemes of GA were designed, include: initial population generating scheme, fitness function design scheme, immunization strategy, adaptive crossover probability and adaptive mutation probability design scheme. By taking the rise time, error integral and overshoot of system response as the performance index, and using genetic algorithm for real-coded of PID parameters, then a group of optimal values were obtained. Simulation results show that the method has a good dynamic performance, superior to the conventional PID controller.