Use Of Multi-objective Genetic Algorithms Research Articles

Rule extraction based on PROMETHEE-assisted multi-objective genetic algorithm for generating interpretable neural networks

Rule extraction from artificial neural networks (ANNs) refers to the process of identifying the underlying decision-making logic used by the network in its predictions. This involves converting the complex, non-linear relationships between inputs and outputs in the network into a set of interpretable rules. The main objective of rule extraction is to enhance the interpretability of decision-making in artificial neural networks (ANNs) and make it more understandable for human experts, especially in critical domains like healthcare, where clear decision-making is essential. Despite the ability of machine learning methods to interpret ANNs, interpreting complex neural networks remains limited, as most studies tend to sacrifice one criterion, related to rule quality, in favor of another. Rule extraction methods applied directly to complex neural networks may produce overly specific rules that are tailored to the training data. Such rules can suffer from overfitting and may not generalize well to unseen data. Also, The use of multiobjective genetic algorithms for rule extraction can result in a large number of rules, making it challenging for decision-makers to interpret them. This study proposes an innovative approach to address challenges in rule extraction from complex artificial neural networks (ANNs) by integrating a multiobjective genetic algorithm with the PROMETHEE method. By integrating these techniques, the study aims to identify meaningful relationships between inputs and outputs of ANNs, generating concise and reliable rules that are ranked based on support, confidence, and lift values. The use of PROMETHEE's rule ranking mechanism enabled the consideration of rule priority during candidate selection, guiding the genetic algorithm towards discovering high-quality rule sets and efficiently exploring the search space. This approach not only enhances the quality of extracted rules but also facilitates efficient decision-making by striking a balance between rule accuracy, fidelity, and comprehensibility, thus contributing to advancing the understanding of complex neural networks. By equally weighting the criteria (support, confidence, and lift) during the rule ranking mechanism, our approach achieved high coverage rates ranging from 94.88 % to 100 % and generated a manageable number of rules that could be easily interpreted by human experts, ranging from 5 to 10.5. These findings demonstrate the potential of our approach to significantly improve decision-making accuracy and interpretability across various real-world applications, making it a promising tool for applications in healthcare, finance, and other fields.

Optimization of the Antifungal Property in a Composite of Polyurethane and Silver Nanoparticles against the Trichophyton rubrum Fungus

This research aims to analyze and optimize the antifungal property of a composite material made of polyurethane (PUR) and silver nanoparticles (AgNPs) against the fungus Trichophyton rubrum to find the optimal parameters that allow the highest inhibition of the fungal growth to be used in healthcare aimed at the population vulnerable to fungal infections, such as people suffering from diabetes mellitus, thus creating an opportunity for the generation of antifungal products for healthcare applications. This study analyzes the effect of three variables on the antifungal properties of the composite material: nanoparticle size, concentration, and the application of an ultrasonic treatment as a method to aid the dispersion of the AgNPs into PUR matrix. The proposed methodology includes tests in accordance with a standar to evaluate the growth inhibition of the fungi on a culture medium. A categorical logistic regression model was adjusted from 23 factorial design with five replicates, which was optimized by the use of multi-objective genetic algorithms. The experimental factors showed a significant effect on the growth inhibition of the fungus, and the optimal levels were determined.

Scalogram-energy based segmentation of surface electromyography signals from swallowing related muscles

Background and Objective: The swallowing is a complex process mediated by the central nervous system, that implies voluntary and involuntary components, including 26 pairs of muscles. Non-invasive strategies, including the surface electromyography (sEMG), have been proposed to evaluate the swallowing. However, such analyses have been mostly descriptive, and the detection of neuromuscular activity has been limited to the visual inspection (VIS). Nonetheless, the VIS lacks reliability since the swallowing related muscles have small size, they are not completely shallow, suffer from cross-talk and have low signal-to-noise ratio (SNR). In this way, we propose a wavelet based method to automatically detect activations in sEMG signals acquired during praxis and swallowing tasks.Methods: The proposed strategy, namely Scalogram-Energy based Segmentation method, was applied on sEMG signals recorded in masseteric, orbicular, supra- and infrahyoid muscles. The method was trained in a database of 35 healthy subjects by the use of multi-objective genetic algorithms and tested via cross-validation, aiming to maximize the F1 score and minimize the timing error between the automatic and VIS related marks. Furthermore, the proposed method was tested in a database of semi-synthetic signals with variable SNR built from signals collected from 10 individuals. Additionally, the method was compared with a double threshold based algorithm as well as with other based on energy and morphological operators.Results: The algorithm achieved a F1 score of 0.82 and almost 13 ms of error in the estimation of onset and offset. Afterwards, we applied the optimized algorithm to a set with semi-synthetic signals with variable SNR, that achieved F1 score of 0.85 for SNR=6 dB and 0.97 for SNR=8 and 10 dB. The mean of the timing error was smaller than 9 ms for SNR=6,8 and 10 dB. The method was also compared with a double threshold based algorithm as well as with other based on energy and morphological operators.Conclusions: The proposed method shown to be useful to automatically analyze the electrophysiological activity associated to praxis and swallowing process. Nonetheless, the obtained results could be extended to other sEMG related applications.

Multi-Objective Genetic Algorithms to Find Most Relevant Volumes of the Brain Related to Alzheimer's Disease and Mild Cognitive Impairment.

Computer-Aided Diagnosis (CAD) represents a relevant instrument to automatically classify between patients with and without Alzheimer's Disease (AD) using several actual imaging techniques. This study analyzes the optimization of volumes of interest (VOIs) to extract three-dimensional (3D) textures from Magnetic Resonance Image (MRI) in order to diagnose AD, Mild Cognitive Impairment converter (MCIc), Mild Cognitive Impairment nonconverter (MCInc) and Normal subjects. A relevant feature of the proposed approach is the use of 3D features instead of traditional two-dimensional (2D) features, by using 3D discrete wavelet transform (3D-DWT) approach for performing feature extraction from T-1 weighted MRI. Due to the high number of coefficients when applying 3D-DWT to each of the VOIs, a feature selection algorithm based on mutual information is used, as is the minimum Redundancy Maximum Relevance (mRMR) algorithm. Region optimization has been performed in order to discover the most relevant regions (VOIs) in the brain with the use of Multi-Objective Genetic Algorithms, being one of the objectives to be optimize the accuracy of the system. The error index of the system is computed by the confusion matrix obtained by the multi-class support vector machine (SVM) classifier. Principal Component Analysis (PCA) is used with the purpose of reducing the number of features to the classifier. The cohort of subjects used in the study consisted of 296 different patients. A first group of 206 patients was used to optimize VOI selection and another group of 90 independent subjects (that did not belong to the first group) was used to test the solutions yielded by the genetic algorithm. The proposed methodology obtains excellent results in multi-class classification achieving accuracies of 94.4% and also extracting significant information on the location of the most relevant points of the brain. This suggests that the proposed method could aid in the research of other neurodegenerative diseases, improving the accuracy of the diagnosis and finding the most relevant regions of the brain associated with them.

EDT, C80 and G Driven Auditorium Design

Acoustical Indexes as described in ISO3382-11, with their optimal values, are not necessarily positively correlated. Some of them are in fact negatively correlated2. This means that when designers try to shape a room in such a way as to obtain optimal EDT and C80 values for a certain position in the room, they face contrasting objectives. This problem is further complicated by the fact that rooms are designed for multiple listening positions with different distances from the sound source and from reflecting surfaces. The distribution of energy in time and space that is required to obtain optimal acoustical values for all listening positions in a room is a complex problem. Multi-Objective Optimization Algorithms can deal with contrasting design objectives and aid the designer by generating or suggesting a series of solutions that are said to be non-dominated from each other. This set of solutions is called the Pareto front, and can be of valuable information for acoustic consultants and architects early in the design process. This paper presents the use of Multi-Objective Genetic Algorithms (MOGAs) for the design of shoebox auditoriums using the EDT, C80 and G indexes as fitness functions to determine the room's main dimensions.

Multi-objective genetic algorithms for solving portfolio optimization problems in the electricity market

The multi-objective portfolio optimization problem is not easy to solve because of (i) challenges from the complexity that arises due to conflicting objectives, (ii) high occurrence of non-dominance of solutions based on the dominance relation, and (iii) optimization solutions that often result in under-diversification. This paper experiments the use of multi-objective genetic algorithms (MOGAs), namely, the non-dominated sorting genetic algorithm II (NSGA-II), strength Pareto evolutionary algorithm II (SPEA-II) and newly proposed compressed objective genetic algorithm II (COGA-II) for solving the portfolio optimization problem for a power generation company (GenCo) faced with different trading choices. To avoid under-diversification, an additional objective to enhance the diversification benefit is proposed alongside with the three original objectives of the mean–variance–skewness (MVS) portfolio framework. The results show that MOGAs have made possible the inclusion of the fourth objective within the optimization framework that produces Pareto fronts that also cover those based on the traditional MVS framework, thereby offering better trade-off solutions while promoting investment diversification benefits for power generation companies.

A novel approach for traffic accidents sanitary resource allocation based on multi-objective genetic algorithms

The development of communication technologies integrated in vehicles allows creating new protocols and applications to improve assistance in traffic accidents. Combining this technology with intelligent systems will permit to automate most of the decisions needed to generate the appropriate sanitary resource sets, thereby reducing the time from the occurrence of the accident to the stabilization and hospitalization of the injured passengers. However, generating the optimal allocation of sanitary resources is not an easy task, since there are several objectives that are mutually exclusive, such as assistance improvement, cost reduction, and balanced resource usage. In this paper, we propose a novel approach for the sanitary resources allocation in traffic accidents. Our approach is based on the use of multi-objective genetic algorithms, and it is able to generate a list of optimal solutions accounting for the most representative factors. The inputs to our model are: (i) the accident notification, which is obtained through vehicular communication systems, and (ii) the severity estimation for the accident, achieved through data mining. We evaluate our approach under a set of vehicular scenarios, and the results show that a memetic version of the NSGA-II algorithm was the most effective method at locating the optimal resource set, while maintaining enough variability in the solutions to allow applying different resource allocation policies.

A Study on the Use of Multiobjective Genetic Algorithms for Classifier Selection in FURIA-based Fuzzy Multiclassifiers

Abstract In a preceding contribution, we conducted a study considering a fuzzy multiclassifier system (MCS) design framework based on Fuzzy Unordered Rule Induction Algorithm (FURIA). It served as the fuzzy rule classification learning algorithm to derive the component classifiers considering bagging and feature selection. In this work, we integrate this approach under the overproduce-and-choose strategy. A state-of-the-art evolutionary multiobjective algorithm, namely NSGA-II, is used to provide a component classifier selection and improve FURIA-based fuzzy MCS. We propose five different fitness functions based on three different optimization criteria, accuracy, complexity, and diversity. Twenty UCI high dimensional datasets were considered in order to conduct the experiments. A combination between accuracy and diversity criteria provided very promising results, becoming competitive with classical MCS learning methods.

Selecting Model Parameter Sets from a Trade-off Surface Generated from the Non-Dominated Sorting Genetic Algorithm-II

There is increasing trend in the use of multi-objective genetic algorithms (GAs) to estimate parameter sets in the calibration of hydrological models. Multi-objective GAs facilitate the evaluation of several model evaluation objectives, and the examination of massive combinations of parameter sets. Typically, the outcome is a set of several equally-accurate parameter sets which make-up a trade-off surface between the objective functions, usually referred to as Pareto set. The Pareto set is a set of incomparable parameter sets as each solution has unique parameter values in parameter space with competing accuracy in the objective function space. As would be required for decision making purposes, a single parameter set is usually chosen to represent the model calibration procedure. An automated framework for choosing a single solution from such a trade-off surface has not been thoroughly investigated in the model calibration literature. As a result, this study has outlined an automated framework using the distribution of solutions in objective space and parameter space to select solutions with unique properties from an incomparable set of solutions. Our Pareto set was generated from the application of Non-dominated Sorting Genetic Algorithm-II (NSGA-II) to calibrate the Soil and Water Assessment Tool (SWAT) for simulations of streamflow in the Fairchild Creek watershed in southern Ontario. Using cluster analysis to evaluate the distribution of solutions in both objective space and parameter space, we developed four auto-selection methods for choosing parameter sets from the trade-off surface to support decision making. Our method generates solutions with unique properties including a representative pathway in parameter space, a basin of attraction (or the center of mass) in objective space, a proximity to the origin in objective space, and a balanced compromise between objective space and parameter space (denoted BCOP). The BCOP method is appealing as it is an equally-weighted compromise for the distribution of solutions in objective space and parameter space. That is, the BCOP solution emphasizes stability in model parameter values and in objective function values—in a way that similarity in parameter space implies similarity in objective space.

Global optimization of an accelerator lattice using multiobjective genetic algorithms

Storage ring lattice design is a highly constrained multiobjective optimization problem. The objectives can include lattice functions or derived quantities like emittance, brightness, or luminosity while simultaneously fulfilling constraints such as linear stability of the lattice. In this paper we explore the use of multiobjective genetic algorithms (MOGA) to find globally optimized lattice settings in a storage ring. Using the Advanced Light Source (ALS) for illustration, three examples of MOGA are shown and analyzed—(i) using three fit parameters to optimize the straight section betatron function and the natural emittance, (ii) using three fit parameters to optimize the photon brightness of bending magnet and insertion device source points in the lattice and (iii) a six parameter fit creating alternating high and low horizontal betatron functions in subsequent straight sections while still minimizing the natural emittance. Making use of one of the main benefits of MOGA, we also study the trade-offs in the optimization objectives between sets of optimal solutions.

Prediction of building's temperature using neural networks models

The use of artificial neural networks in various applications related with energy management in buildings has been increasing significantly over the recent years. In this paper the design of inside air temperature predictive neural network models, to be used for predictive control of air-conditioned systems, is discussed. The use of multi-objective genetic algorithms for designing off-line radial basis function neural network models is detailed. The performance of these data-driven models is compared, favourably, with a multi-node physically based model. Climate and environmental data from a secondary school located in the south of Portugal, collected by a remote data acquisition system, are used to generate the models. By using a sliding window adaptive methodology, the good results obtained off-line are extended throughout the whole year. The use of long-range predictive models for air-conditioning systems control is demonstrated, in simulations, achieving a good temperature regulation with important energy savings.

System optimization by multiobjective genetic algorithms and analysis of the coupling between variables, constraints and objectives

PurposeThis paper presents a methodology based on Multiobjective Genetic Algorithms (MOGAs) for the design of electrical engineering systems. MOGAs allow one to optimize multiple heterogeneous criteria in complex systems, but also simplify couplings and sensitivity analysis by determining the evolution of design variables along the Pareto‐optimal front.Design/methodology/approachTo illustrate the use of MOGAs in electrical engineering, the optimal design of an electromechanical system has been investigated. A rather simplified case study dealing with the optimal dimensioning of an inverter – permanent magnet motor – reducer – load association is carried out to demonstrate the interest of the approach. The purpose is to simultaneously minimize two objectives: the global losses and the mass of the system. The system model is described by analytical model and we use the MOGA called NSGA‐II.FindingsFrom the extraction of Pareto‐optimal solutions, MOGAs facilitate the investigation of parametric sensitivity and the analysis of couplings in the system. Through a simple but typical academic problem dealing with the optimal dimensioning of a inverter – permanent magnet motor – reducer – load association, it has been shown that this multiobjective a posteriori approach could offer interesting outlooks in the global optimization and design of complex heterogeneous systems. The final choice between all Pareto‐optimal configurations can be a posteriori done in relation to other issues which have not been considered in the optimization process. In this paper, we illustrate this point by considering the cogging torque for the final decision.Originality/valueWe have proposed an original quantitative methodology based on correlation coefficients to characterize the system interactions.

Enhancing the electrical performance of a solid oxide fuel cell using multiobjective genetic algorithms

Several technologies are being used in distributed generation applications with variable degree of success. Among those are: wind turbines, small-scale hydropower plants, biomass, microturbines, photovoltaic arrays, and fuel cells. A fuel cell is a device that converts the chemical energy of fuel to electric energy. New improvements in the fuel cell technology significantly meliorated the technical characteristics of this technology. Environmental friendliness, practically noise free operation, and very high efficiency make fuel cells a challenger on the future electricity markets. The connection between the fuel cell and the load is through an inverter using Pulse Width Modulation. Subsequently, the fuel cell produces a significant amount of harmonics. This paper describes the use of multiobjective genetic algorithms in the design of a fuzzy logic control system for a solid oxide fuel cell.

Use of multiobjective genetic algorithms to optimize inter-vehicle active suspensions

This paper studies inter-vehicle active suspensions for railway vehicles and presents an optimization process for the design of vertical active suspension controllers using multiobjective genetic algorithms. A three-vehicle train set is used in the study and two active control schemes are considered primarily to provide the best improvement in the passenger ride quality. The first scheme uses only actuators placed between adjacent vehicles while the second adds two actuators between bogie and vehicle body at either end of the train set in addition to the inter-vehicle actuators. The development of the control laws is assisted by the use of genetic algorithms to achieve the ‘best’ compromise of different design criteria, especially that between the ride quality and the suspension deflections. The study shows that, when the control laws for the proposed active schemes are optimized, a significant improvement in the vertical ride quality on random tracks is obtained and in the mean time the suspension deflections can be kept within their allowed clearance when the vehicles run on to a gradient.

Exploring use of multiobjective genetic algorithms for constrained process scheduling

Abstract Chemical processing offers constrained, complex scheduling problems (Nott and Lee, 1997). Genetic algorithms (GAs) are evolutionary-based optimisation tools (Holland, 1975), suited to schedule optimisation search, (Davis, 1985). Multi-objective GAs (MOGA) allow multiple objective optimisation (Fonseca and Fleming, 1998), as well as a technique for constraint handling, and the exploration of the inherent trade-offs between goals and constraints. A batch process problem is introduced, and a MOGA is applied to a plant model. The MOGA acts as a flexible optimisation tool, incorporating the framework of the ISA standard S88 (Love and Bunch, 1998) and allowing the user to specify plant variables. This paper explores the MOGA as an optimisation and constraint satisfaction method.

Multiobjective gas turbine engine controller design using genetic algorithms

This paper describes the use of multiobjective genetic algorithms (MOGAs) in the design of a multivariable control system for a gas turbine engine. The mechanisms employed to facilitate multiobjective search with the genetic algorithm are described with the aid of an example. It is shown that the MOGA confers a number of advantages over conventional multiobjective optimization methods by evolving a family of Pareto-optimal solutions rather than a single solution estimate. This allows the engineer to examine the trade-offs between the different design objectives and configurations during the course of an optimization. In addition, the paper demonstrates how the genetic algorithm can be used to search in both controller structure and parameter space thereby offering a potentially more general approach to optimization in controller design than traditional numerical methods. While the example in the paper deals with control system design, the approach described can be expected to be applicable to more general problems in the fields of computer aided design (CAD) and computer aided engineering (CAE).