Skip to main content

Aurora Pozo

Followers

22

Following

1

Public Views

Harvard University

Jordi Vallverdú

Universitat Autònoma de Barcelona

Amrita Vishwa Vidyapeetham

Armando Marques-Guedes

UNL - New University of Lisbon

Geisinger Health System

Near East University

Oxford Brookes University

Roshan Chitrakar

Nepal College of Information Technology

Graduate Center of the City University of New York

University of Technology Sydney

Interests

Uploads

Papers by Aurora Pozo

Evolving convolutional neural networks through grammatical evolution

Proceedings of the Genetic and Evolutionary Computation Conference Companion, Jul 13, 2019

The use of Convolutional Neural Networks (CNNs) have proven to be a solid approach often used to ... more The use of Convolutional Neural Networks (CNNs) have proven to be a solid approach often used to solve many Machine Learning problems, such as image classification and natural language processing tasks. The manual design of CNNs is a hard task, due to the high number of configurable parameters and possible configurations. Recent studies around the automatic design of CNNs have shown positive results. In this work, we propose to explore the design of CNN architectures through the use of Grammatical Evolution (GE), where a BNF grammar is used to define the CNN components and structural rules. Experiments were performed using the MNIST and CIFAR-10 datasets. The obtained results show that the presented approach achieved competitive results and maintaining relatively small architectures when compared with similar state-of-the-art approaches.

An empirical comparison of distance/similarity measures for Natural Language Processing

Anais do XVI Encontro Nacional de Inteligência Artificial e Computacional (ENIAC 2019), 2020

Text Classification is one of the tasks of Natural Language Processing (NLP). In this area, Graph... more Text Classification is one of the tasks of Natural Language Processing (NLP). In this area, Graph Convolutional Networks (GCN) has achieved values higher than CNN's and other related models. For GCN, the metric that defines the correlation between words in a vector space plays a crucial role in the classification because it determines the weight of the edges between two words (represented by nodes in the graph). In this study, we empirically investigated the impact of thirteen measures of distance/similarity. A representation was built for each document using word embedding from word2vec model. Also, a graph-based representation of five dataset was created for each measure analyzed, where each word is a node in the graph, and each edge is weighted by distance/similarity between words. Finally, each model was run in a simple graph neural network. The results show that, concerning text classification, there is no statistical difference between the analyzed metrics and the Graph Co...

Not all PBILs are the same: Unveiling the different learning mechanisms of PBIL variants

Applied Soft Computing, 2017

Model-based optimization using probabilistic modeling of the search space is one of the areas whe... more Model-based optimization using probabilistic modeling of the search space is one of the areas where research on evolutionary algorithms (EAs) has considerably advanced in recent years. The populationbased incremental algorithm (PBIL) is one of the first algorithms of its kind and it has been extensively applied to many optimization problems. In this paper we show that the different applications of PBIL reported in the literature correspond, in fact, to two essentially different algorithms, which are defined by the way the learning step is implemented. We analytically and empirically study the impact of the learning method on the search behavior of the algorithm. As a result of our research, we show examples in which the choice of a PBIL variant can produce qualitatively different outputs of the search process.

Discovering motifs and genomic patterns with SMT: a high-performance data structure for counting kmers

bioRxiv (Cold Spring Harbor Laboratory), Apr 2, 2023

The search for conserved motifs in DNA sequences is an important problem in bioinformatics. The g... more The search for conserved motifs in DNA sequences is an important problem in bioinformatics. The growing availability of large-scale genomic data poses significant challenges for computational biology, particularly in terms of efficiency in analysis, kmer identification, and noise presence. The detection of conserved motifs and patterns in DNA sequences is crucial for understanding gene functions and regulations. Therefore, it is essential to develop a data structure that can handle these large volumes of information and provide accurate and fast results. Results: We present SMT, an innovative tool designed to efficiently store and count kmers, optimizing memory usage and computation time. The application of SMT has also proven effective in discovering motifs in noisy datasets, allowing the identification of conserved regions in sequences. Furthermore, SMT enables exact searches in constant time and recovers the most abundant k-mers, as well as performs approximate searches in linear time to find fragments with up to d mutations. This approach facilitates large-scale data analysis and provides important insights into the conserved properties of biological sequences. The application of SMT in motif discovery demonstrates its potential to drive research in bioinformatics and genomics. Supplementary data and results are available to provide additional information and support the conclusions presented in this work.

A multi-objective and evolutionary hyper-heuristic applied to the Integration and Test Order Problem

Applied Soft Computing, Jul 1, 2017

The field of Search-Based Software Engineering (SBSE) has widely utilized Multi-Objective Evoluti... more The field of Search-Based Software Engineering (SBSE) has widely utilized Multi-Objective Evolutionary Algorithms (MOEAs) to solve complex software engineering problems. However, the use of such algorithms can be a hard task for the software engineer, mainly due to the significant range of parameter and algorithm choices. To help in this task, the use of Hyper-heuristics is recommended. Hyper-heuristics can select or generate low-level heuristics while optimization algorithms are executed, and thus can be generically applied. Despite their benefits, we find only a few works using hyper-heuristics in the SBSE field. Considering this fact, we describe HITO, a Hyper-heuristic for the Integration and Test Order problem, to adaptively select search operators while MOEAs are executed using one of the selection methods: Choice Function and Multi-Armed Bandit. The experimental results show that HITO can outperform the traditional MOEAs NSGA-II and MOEA/DD. HITO is also a generic algorithm, since the user does not need to select crossover and mutation operators, nor adjust their parameters.

Towards a Better Understanding of Heuristic Approaches Applied to the Biological Motif Discovery

Springer eBooks, 2022

A Rule Learning Multiobjective Particle Swarm Optimization

IEEE Latin America Transactions, Aug 1, 2009

Multiobjective Metaheuristics (MOMH) permit to conceive a complete novel approach to induce class... more Multiobjective Metaheuristics (MOMH) permit to conceive a complete novel approach to induce classifiers. In the Rule Learning problem, the use of MOMH permit that the properties of the rules can be expressed in different objectives, and then the algorithm finds these rules in an unique run by exploring Pareto dominance concepts. This work describes a Multiobjective Particle Swarm Optimization (MOPSO) algorithm that handles with numerical and discrete attributes. The algorithm is evaluated by using the area under ROC curve and the approximation sets produced by the algorithm are also analyzed following Multiobjective methodology.

Genetic algorithm restricted by tabu lists in data mining

The present work shows an implementation of a genetic algorithm (GA) integrated with tabu lists t... more The present work shows an implementation of a genetic algorithm (GA) integrated with tabu lists to generate a classifier tool for a Data Mining task. The choice of GAs paradigm is partially justified by its great capacity in dealing with noise, invalid or inexact data, and its easy adaptation to different domains of data. The GA algorithm uses Tabu List to restrict the selection process. This restriction allows the creation of set of potential rules for the classifier tool. This strategy was proposed in , for multimodal and multiobjective function optimization and represents an alternative to sharing methods. In this work, the behavior of this approach in Data Mining task was analyzed. Experiments were performed on five databases and results were compared with 34 other classifying algorithms. After that, noise was added to the databases and a new set of experiments was performed. The results show that the algorithm proposed herein is efficient and robust. And the strategy used to maintain the diversity was considered valid, since the algorithm was able to keep its accuracy in categorization even for smaller populations.

A Meta-Learning Algorithm Selection Approach for the Quadratic Assignment Problem

The Quadratic Assignment Problem is a classical hard combinatorial optimization problem for which... more The Quadratic Assignment Problem is a classical hard combinatorial optimization problem for which many strategies have been proposed. However, as several empirical studies have shown, there is not a single best strategy for all cases, meaning that they perform differently according to the problem instance. This can be addressed under the formulation of an Algorithm Selection Problem and, to solve it, we propose a cascade classifying system, following the concepts of meta-learning. After performing data cleansing and dimensionality reduction on the datasets, our cascade scheme was able to show very promising results, both in overall accuracy and in individual classes performances.

On the use of fitness landscape features in meta-learning based algorithm selection for the quadratic assignment problem

Theoretical Computer Science, 2020

Grammatical Evolution for the Multi-Objective Integration and Test Order Problem

Search techniques have been successfully applied for solving different software testing problems.... more Search techniques have been successfully applied for solving different software testing problems. However, choosing, implementing and configuring a search technique can be hard tasks. To reduce efforts spent in such tasks, this paper presents an offline hyper-heuristic named GEMOITO, based on Grammatical Evolution (GE). The goal is to automatically generate a Multi-Objective Evolutionary Algorithm (MOEA) to solve the Integration and Test Order (ITO) problem. The MOEAs are distinguished by components and parameters values, described by a grammar. The proposed hyper-heuristic is compared to conventional MOEAs and to a selection hyper-heuristic used in related work. Results show that GEMOITO can generate MOEAs that are statistically better or equivalent to the compared algorithms.

Automated design of hyper-heuristics components to solve the PSP problem with HP model

The Protein Structure Prediction (PSP) problem is one of the modern most challenging problems fro... more The Protein Structure Prediction (PSP) problem is one of the modern most challenging problems from science. Simplified protein models are usually applied to simulate and study some characteristics of the protein folding process. Hence, many heuristic strategies have been applied in order to find simplified protein structures in which the protein configuration has the minimal energy. However, these strategies have difficulties in finding the optimal solutions to the longer sequences of amino-acids, due to the complexity of the problem and the huge amount of local optima. Hyper heuristics have proved to be useful in this type of context since they try to combine different heuristics strengths into a single framework. However, there is lack of work addressing the automated design of hyper-heuristics components. This paper proposes GEHyPSP, an approach which aims to achieve generation, through grammatical evolution, of selection mechanisms and acceptance criteria for a hyper-heuristic framework applied to PSP problem. We investigate the strengths and weaknesses of our approach on a benchmark of simplified protein models. GEHyPSP was able to reach the best known results for 7 instances from 11 that composed the benchmark set used to evaluate the approach.

An Analysis of the Admissibility of the Objective Functions Applied in Evolutionary Multi-objective Clustering

arXiv (Cornell University), Jun 19, 2022

A variety of clustering criteria has been applied as an objective function in Evolutionary Multi-... more A variety of clustering criteria has been applied as an objective function in Evolutionary Multi-Objective Clustering approaches (EMOCs). However, most EMOCs do not provide detailed analysis regarding the choice and usage of the objective functions. Aiming to support a better choice and definition of the objectives in the EMOCs, this paper proposes an analysis of the admissibility of the clustering criteria in evolutionary optimization by examining the search direction and its potential in finding optimal results. As a result, we demonstrate how the admissibility of the objective functions can influence the optimization. Furthermore, we provide insights regarding the combinations and usage of the clustering criteria in the EMOCs.

Ant-Miner Specializations to tackle Imbalanced Data Sets

A data set is named imbalanced when the classes have not approximately equally representations. C... more A data set is named imbalanced when the classes have not approximately equally representations. Classification algorithms are sensitive of this imbalance and tend to valorize the majority classes and ignore the minority classes, which is a problem when the minority classes are the classes of interest. In this paper we propose two specializations to an efficient and robust classification algorithm inspired by ACO metaheuristic called Ant-Miner. These specializations modify how rules are constructed and evaluated. We compare the results with standard Ant-Miner and C4.5 algorithm. The results show that the proposed algorithms are competitive, finding rules for the minority classes and improve the simplicity of the discovered rule list.

Creating deep neural networks for text classification tasks using grammar genetic programming

Applied Soft Computing, Mar 1, 2023

Evolving Connection Weights of Artificial Neural Network Using a Multi-Objective Approach with Application to Class Prediction

Studies in computational intelligence, Oct 2, 2016

In Artificial Neural Network (ANN), the selection of connection weights is a key issue and Geneti... more In Artificial Neural Network (ANN), the selection of connection weights is a key issue and Genetic and Evolution Strategies have been found to be promising algorithms to solve this important task. Motivated by that, this study investigates the applicability of using two novel Multi-Objective Evolutionary Algorithms (MOEA): Speed constrained Multi-Objective Particle Swarm Optimization (SMPSO) and Multi-Objective Differential Evolution Algorithm based on Decomposition with Dynamical Resource Allocation (MOEA/D-DE-DRA). ANNs are training to learn data classification using sensibility and specificity for different UCI databases. The results are compared using the Hypervolume as quality indicator and statistical test.

Automatic Design of Deep Neural Networks Applied to Image Segmentation Problems

Lecture Notes in Computer Science, 2021

A study on auto-configuration of Multi-Objective Particle Swarm Optimization Algorithm

Researches point out to the importance of automatic design of multi-objective evolutionary algori... more Researches point out to the importance of automatic design of multi-objective evolutionary algorithms. Because in general, algorithms automatically designed outperform traditional multi-objective evolutionary algorithms from the literature. Nevertheless, until fairly recently, most of the researches have been focused on a small group of algorithms, often based on evolutionary algorithms. On the other hand, mono-objective Particle Swarm Optimization algorithm (PSO) have been widely used due to its flexibility and competitive results in different applications. Besides, as PSO performance depends on different aspects of design like the velocity equation, its automatic design has been targeted by many researches with encouraging results. Motivated by these issues, this work studies the automatic design of Multi-Objective Particle Swarm Optimization (MOPSO). A framework that uses a context-free grammar to guide the design of the algorithms is implemented. The framework includes a set of parameters and components of different MOPSOs, and two design algorithms: Grammatical Evolution (GE) and Iterated Racing (IRACE). Evaluation results are presented, comparing MOPSOs generated by both design algorithms. Furthermore, the generated MOPSOs are compared to the Speed-constrained MOPSO (SMPSO), a well-known algorithm using a set of Multi-Objective problems, quality indicators and statistical tests.

Uma proposta para Federated Learning em Cibersegurança

Anais da XXIII Escola Regional de Alto Desempenho da Região Sul (ERAD-RS 2023)

Neste artigo, propõem-se dois métodos de votação para Ensemble Learning em Federated Learning e C... more

Estabelecendo Sequências de Teste de Integração de Classes: Um Estudo Comparativo da Aplicação de Três Algoritmos Evolutivos Multiobjetivos

Anais do XII Workshop de Testes e Tolerância a Falhas (WTF 2011)

Com o intuito de minimizar os custos associados à criação de stubs, durante o teste de integração... more Com o intuito de minimizar os custos associados à criação de stubs, durante o teste de integração de sistemas orientados a objetos é preciso determinar a melhor ordem de integração e teste das classes. Em alguns estudos recentes, Algoritmos Evolutivos Multiobjetivos (MOEA) têm resolvido eficientemente este problema porque eles consideram vários fatores que afetam o processo de construção de stubs. Geralmente esses fatores são conflitantes e por isto, nem sempre é possível obter uma solução (ordem) única. Dentre os trabalhos publicados, os MOEAs NSGA-II e SPEA2 alcançaram bons resultados para o problema em questão. No entanto, existem outros MOEAs que não foram aplicados neste contexto e que podem melhorar o desempenho desses dois algoritmos. Com este objetivo, o presente trabalho explora o uso do algoritmo PAES neste contexto. Foi realizado um experimento com quatro sistemas reais no qual o PAES foi comparado com o NSGA-II e o SPEA2. Os resultados mostram que ele consegue superar os...

Evolving convolutional neural networks through grammatical evolution

Proceedings of the Genetic and Evolutionary Computation Conference Companion, Jul 13, 2019

The use of Convolutional Neural Networks (CNNs) have proven to be a solid approach often used to ... more The use of Convolutional Neural Networks (CNNs) have proven to be a solid approach often used to solve many Machine Learning problems, such as image classification and natural language processing tasks. The manual design of CNNs is a hard task, due to the high number of configurable parameters and possible configurations. Recent studies around the automatic design of CNNs have shown positive results. In this work, we propose to explore the design of CNN architectures through the use of Grammatical Evolution (GE), where a BNF grammar is used to define the CNN components and structural rules. Experiments were performed using the MNIST and CIFAR-10 datasets. The obtained results show that the presented approach achieved competitive results and maintaining relatively small architectures when compared with similar state-of-the-art approaches.

An empirical comparison of distance/similarity measures for Natural Language Processing

Anais do XVI Encontro Nacional de Inteligência Artificial e Computacional (ENIAC 2019), 2020

Text Classification is one of the tasks of Natural Language Processing (NLP). In this area, Graph... more Text Classification is one of the tasks of Natural Language Processing (NLP). In this area, Graph Convolutional Networks (GCN) has achieved values higher than CNN's and other related models. For GCN, the metric that defines the correlation between words in a vector space plays a crucial role in the classification because it determines the weight of the edges between two words (represented by nodes in the graph). In this study, we empirically investigated the impact of thirteen measures of distance/similarity. A representation was built for each document using word embedding from word2vec model. Also, a graph-based representation of five dataset was created for each measure analyzed, where each word is a node in the graph, and each edge is weighted by distance/similarity between words. Finally, each model was run in a simple graph neural network. The results show that, concerning text classification, there is no statistical difference between the analyzed metrics and the Graph Co...

Not all PBILs are the same: Unveiling the different learning mechanisms of PBIL variants

Applied Soft Computing, 2017

Model-based optimization using probabilistic modeling of the search space is one of the areas whe... more Model-based optimization using probabilistic modeling of the search space is one of the areas where research on evolutionary algorithms (EAs) has considerably advanced in recent years. The populationbased incremental algorithm (PBIL) is one of the first algorithms of its kind and it has been extensively applied to many optimization problems. In this paper we show that the different applications of PBIL reported in the literature correspond, in fact, to two essentially different algorithms, which are defined by the way the learning step is implemented. We analytically and empirically study the impact of the learning method on the search behavior of the algorithm. As a result of our research, we show examples in which the choice of a PBIL variant can produce qualitatively different outputs of the search process.

Discovering motifs and genomic patterns with SMT: a high-performance data structure for counting kmers

bioRxiv (Cold Spring Harbor Laboratory), Apr 2, 2023

The search for conserved motifs in DNA sequences is an important problem in bioinformatics. The g... more The search for conserved motifs in DNA sequences is an important problem in bioinformatics. The growing availability of large-scale genomic data poses significant challenges for computational biology, particularly in terms of efficiency in analysis, kmer identification, and noise presence. The detection of conserved motifs and patterns in DNA sequences is crucial for understanding gene functions and regulations. Therefore, it is essential to develop a data structure that can handle these large volumes of information and provide accurate and fast results. Results: We present SMT, an innovative tool designed to efficiently store and count kmers, optimizing memory usage and computation time. The application of SMT has also proven effective in discovering motifs in noisy datasets, allowing the identification of conserved regions in sequences. Furthermore, SMT enables exact searches in constant time and recovers the most abundant k-mers, as well as performs approximate searches in linear time to find fragments with up to d mutations. This approach facilitates large-scale data analysis and provides important insights into the conserved properties of biological sequences. The application of SMT in motif discovery demonstrates its potential to drive research in bioinformatics and genomics. Supplementary data and results are available to provide additional information and support the conclusions presented in this work.

A multi-objective and evolutionary hyper-heuristic applied to the Integration and Test Order Problem

Applied Soft Computing, Jul 1, 2017

The field of Search-Based Software Engineering (SBSE) has widely utilized Multi-Objective Evoluti... more The field of Search-Based Software Engineering (SBSE) has widely utilized Multi-Objective Evolutionary Algorithms (MOEAs) to solve complex software engineering problems. However, the use of such algorithms can be a hard task for the software engineer, mainly due to the significant range of parameter and algorithm choices. To help in this task, the use of Hyper-heuristics is recommended. Hyper-heuristics can select or generate low-level heuristics while optimization algorithms are executed, and thus can be generically applied. Despite their benefits, we find only a few works using hyper-heuristics in the SBSE field. Considering this fact, we describe HITO, a Hyper-heuristic for the Integration and Test Order problem, to adaptively select search operators while MOEAs are executed using one of the selection methods: Choice Function and Multi-Armed Bandit. The experimental results show that HITO can outperform the traditional MOEAs NSGA-II and MOEA/DD. HITO is also a generic algorithm, since the user does not need to select crossover and mutation operators, nor adjust their parameters.

Towards a Better Understanding of Heuristic Approaches Applied to the Biological Motif Discovery

Springer eBooks, 2022

A Rule Learning Multiobjective Particle Swarm Optimization

IEEE Latin America Transactions, Aug 1, 2009

Multiobjective Metaheuristics (MOMH) permit to conceive a complete novel approach to induce class... more Multiobjective Metaheuristics (MOMH) permit to conceive a complete novel approach to induce classifiers. In the Rule Learning problem, the use of MOMH permit that the properties of the rules can be expressed in different objectives, and then the algorithm finds these rules in an unique run by exploring Pareto dominance concepts. This work describes a Multiobjective Particle Swarm Optimization (MOPSO) algorithm that handles with numerical and discrete attributes. The algorithm is evaluated by using the area under ROC curve and the approximation sets produced by the algorithm are also analyzed following Multiobjective methodology.

Genetic algorithm restricted by tabu lists in data mining

The present work shows an implementation of a genetic algorithm (GA) integrated with tabu lists t... more The present work shows an implementation of a genetic algorithm (GA) integrated with tabu lists to generate a classifier tool for a Data Mining task. The choice of GAs paradigm is partially justified by its great capacity in dealing with noise, invalid or inexact data, and its easy adaptation to different domains of data. The GA algorithm uses Tabu List to restrict the selection process. This restriction allows the creation of set of potential rules for the classifier tool. This strategy was proposed in , for multimodal and multiobjective function optimization and represents an alternative to sharing methods. In this work, the behavior of this approach in Data Mining task was analyzed. Experiments were performed on five databases and results were compared with 34 other classifying algorithms. After that, noise was added to the databases and a new set of experiments was performed. The results show that the algorithm proposed herein is efficient and robust. And the strategy used to maintain the diversity was considered valid, since the algorithm was able to keep its accuracy in categorization even for smaller populations.

A Meta-Learning Algorithm Selection Approach for the Quadratic Assignment Problem

The Quadratic Assignment Problem is a classical hard combinatorial optimization problem for which... more The Quadratic Assignment Problem is a classical hard combinatorial optimization problem for which many strategies have been proposed. However, as several empirical studies have shown, there is not a single best strategy for all cases, meaning that they perform differently according to the problem instance. This can be addressed under the formulation of an Algorithm Selection Problem and, to solve it, we propose a cascade classifying system, following the concepts of meta-learning. After performing data cleansing and dimensionality reduction on the datasets, our cascade scheme was able to show very promising results, both in overall accuracy and in individual classes performances.

On the use of fitness landscape features in meta-learning based algorithm selection for the quadratic assignment problem

Theoretical Computer Science, 2020

Grammatical Evolution for the Multi-Objective Integration and Test Order Problem

Search techniques have been successfully applied for solving different software testing problems.... more Search techniques have been successfully applied for solving different software testing problems. However, choosing, implementing and configuring a search technique can be hard tasks. To reduce efforts spent in such tasks, this paper presents an offline hyper-heuristic named GEMOITO, based on Grammatical Evolution (GE). The goal is to automatically generate a Multi-Objective Evolutionary Algorithm (MOEA) to solve the Integration and Test Order (ITO) problem. The MOEAs are distinguished by components and parameters values, described by a grammar. The proposed hyper-heuristic is compared to conventional MOEAs and to a selection hyper-heuristic used in related work. Results show that GEMOITO can generate MOEAs that are statistically better or equivalent to the compared algorithms.

Automated design of hyper-heuristics components to solve the PSP problem with HP model

The Protein Structure Prediction (PSP) problem is one of the modern most challenging problems fro... more The Protein Structure Prediction (PSP) problem is one of the modern most challenging problems from science. Simplified protein models are usually applied to simulate and study some characteristics of the protein folding process. Hence, many heuristic strategies have been applied in order to find simplified protein structures in which the protein configuration has the minimal energy. However, these strategies have difficulties in finding the optimal solutions to the longer sequences of amino-acids, due to the complexity of the problem and the huge amount of local optima. Hyper heuristics have proved to be useful in this type of context since they try to combine different heuristics strengths into a single framework. However, there is lack of work addressing the automated design of hyper-heuristics components. This paper proposes GEHyPSP, an approach which aims to achieve generation, through grammatical evolution, of selection mechanisms and acceptance criteria for a hyper-heuristic framework applied to PSP problem. We investigate the strengths and weaknesses of our approach on a benchmark of simplified protein models. GEHyPSP was able to reach the best known results for 7 instances from 11 that composed the benchmark set used to evaluate the approach.

An Analysis of the Admissibility of the Objective Functions Applied in Evolutionary Multi-objective Clustering

arXiv (Cornell University), Jun 19, 2022

A variety of clustering criteria has been applied as an objective function in Evolutionary Multi-... more A variety of clustering criteria has been applied as an objective function in Evolutionary Multi-Objective Clustering approaches (EMOCs). However, most EMOCs do not provide detailed analysis regarding the choice and usage of the objective functions. Aiming to support a better choice and definition of the objectives in the EMOCs, this paper proposes an analysis of the admissibility of the clustering criteria in evolutionary optimization by examining the search direction and its potential in finding optimal results. As a result, we demonstrate how the admissibility of the objective functions can influence the optimization. Furthermore, we provide insights regarding the combinations and usage of the clustering criteria in the EMOCs.

Ant-Miner Specializations to tackle Imbalanced Data Sets

A data set is named imbalanced when the classes have not approximately equally representations. C... more A data set is named imbalanced when the classes have not approximately equally representations. Classification algorithms are sensitive of this imbalance and tend to valorize the majority classes and ignore the minority classes, which is a problem when the minority classes are the classes of interest. In this paper we propose two specializations to an efficient and robust classification algorithm inspired by ACO metaheuristic called Ant-Miner. These specializations modify how rules are constructed and evaluated. We compare the results with standard Ant-Miner and C4.5 algorithm. The results show that the proposed algorithms are competitive, finding rules for the minority classes and improve the simplicity of the discovered rule list.

Creating deep neural networks for text classification tasks using grammar genetic programming

Applied Soft Computing, Mar 1, 2023

Evolving Connection Weights of Artificial Neural Network Using a Multi-Objective Approach with Application to Class Prediction

Studies in computational intelligence, Oct 2, 2016

In Artificial Neural Network (ANN), the selection of connection weights is a key issue and Geneti... more In Artificial Neural Network (ANN), the selection of connection weights is a key issue and Genetic and Evolution Strategies have been found to be promising algorithms to solve this important task. Motivated by that, this study investigates the applicability of using two novel Multi-Objective Evolutionary Algorithms (MOEA): Speed constrained Multi-Objective Particle Swarm Optimization (SMPSO) and Multi-Objective Differential Evolution Algorithm based on Decomposition with Dynamical Resource Allocation (MOEA/D-DE-DRA). ANNs are training to learn data classification using sensibility and specificity for different UCI databases. The results are compared using the Hypervolume as quality indicator and statistical test.

Automatic Design of Deep Neural Networks Applied to Image Segmentation Problems

Lecture Notes in Computer Science, 2021

A study on auto-configuration of Multi-Objective Particle Swarm Optimization Algorithm

Researches point out to the importance of automatic design of multi-objective evolutionary algori... more Researches point out to the importance of automatic design of multi-objective evolutionary algorithms. Because in general, algorithms automatically designed outperform traditional multi-objective evolutionary algorithms from the literature. Nevertheless, until fairly recently, most of the researches have been focused on a small group of algorithms, often based on evolutionary algorithms. On the other hand, mono-objective Particle Swarm Optimization algorithm (PSO) have been widely used due to its flexibility and competitive results in different applications. Besides, as PSO performance depends on different aspects of design like the velocity equation, its automatic design has been targeted by many researches with encouraging results. Motivated by these issues, this work studies the automatic design of Multi-Objective Particle Swarm Optimization (MOPSO). A framework that uses a context-free grammar to guide the design of the algorithms is implemented. The framework includes a set of parameters and components of different MOPSOs, and two design algorithms: Grammatical Evolution (GE) and Iterated Racing (IRACE). Evaluation results are presented, comparing MOPSOs generated by both design algorithms. Furthermore, the generated MOPSOs are compared to the Speed-constrained MOPSO (SMPSO), a well-known algorithm using a set of Multi-Objective problems, quality indicators and statistical tests.

Uma proposta para Federated Learning em Cibersegurança

Anais da XXIII Escola Regional de Alto Desempenho da Região Sul (ERAD-RS 2023)

Neste artigo, propõem-se dois métodos de votação para Ensemble Learning em Federated Learning e C... more

Estabelecendo Sequências de Teste de Integração de Classes: Um Estudo Comparativo da Aplicação de Três Algoritmos Evolutivos Multiobjetivos

Anais do XII Workshop de Testes e Tolerância a Falhas (WTF 2011)

Com o intuito de minimizar os custos associados à criação de stubs, durante o teste de integração... more Com o intuito de minimizar os custos associados à criação de stubs, durante o teste de integração de sistemas orientados a objetos é preciso determinar a melhor ordem de integração e teste das classes. Em alguns estudos recentes, Algoritmos Evolutivos Multiobjetivos (MOEA) têm resolvido eficientemente este problema porque eles consideram vários fatores que afetam o processo de construção de stubs. Geralmente esses fatores são conflitantes e por isto, nem sempre é possível obter uma solução (ordem) única. Dentre os trabalhos publicados, os MOEAs NSGA-II e SPEA2 alcançaram bons resultados para o problema em questão. No entanto, existem outros MOEAs que não foram aplicados neste contexto e que podem melhorar o desempenho desses dois algoritmos. Com este objetivo, o presente trabalho explora o uso do algoritmo PAES neste contexto. Foi realizado um experimento com quatro sistemas reais no qual o PAES foi comparado com o NSGA-II e o SPEA2. Os resultados mostram que ele consegue superar os...