Nuno Lourenço

Unveiling the properties of structured grammatical evolution

Structured grammatical evolution (SGE) is a new genotypic representation for grammatical evolution (GE). It comprises a hierarchical organization of the genes, where each locus is explicitly linked to a non-terminal of the grammar being used. This one-to-one correspondence ensures that the modification of a gene does not affect the derivation options of other non-terminals. We present a comprehensive set of optimization results obtained with problems from three different categories: symbolic regression, path finding, and predictive modeling. In most of the situations SGE outperforms standard GE, confirming the effectiveness of the new representation. To understand the reasons for SGE enhanced performance, we scrutinize its main features. We rely on a set of static measures to model the interactions between the representation and variation operators and assess how they influence the interplay between the genotype-phenotype spaces. The study reveals that the structured organization of SGE promotes an increased locality and is less redundant than standard GE, thus fostering an effective exploration of the search space.

Evolving evolutionary algorithms

This paper proposes a Grammatical Evolution framework to the automatic design of Evolutionary Algorithms. We define a grammar that has the ability to combine components regularly appearing in existing evolutionary algorithms, aiming to achieve novel and fully functional optimization methods. The problem of the Royal Road Functions is used to assess the capacity of the framework to evolve algorithms. Results show that the computational system is able to evolve simple evolutionary algorithms that can effectively solve Royal Road instances. Moreover, some unusual design solutions, competitive with standard approaches, are also proposed by the grammatical evolution framework.

SGE: A Structured Representation for Grammatical Evolution

This paper introduces Structured Grammatical Evolution, a new genotypic representation for Grammatical Evolution, where each gene is explicitly linked to a non-terminal of the grammar being used. This one-to-one correspondence ensures that the modification of a gene does not affect the derivation options of other non-terminals, thereby increasing locality. The performance of the new representation is accessed on a set of benchmark problems. The results obtained confirm the effectiveness of the proposed approach, as it is able to outperform standard grammatical evolution on all selected optimization problems.

The importance of the learning conditions in hyper-heuristics

Evolutionary Algorithms are problem solvers inspired by nature. The effectiveness of these methods on a specific task usually depends on a non trivial manual crafting of their main components and settings. Hyper-Heuristics is a recent area of research that aims to overcome this limitation by advocating the automation of the optimization algorithm design task. In this paper, we describe a Grammatical Evolution framework to automatically design evolutionary algorithms to solve the knapsack problem. We focus our attention on the evaluation of solutions that are iteratively generated by the Hyper-Heuristic. When learning optimization strategies, the hyper-method must evaluate promising candidates by executing them. However, running an evolutionary algorithm is an expensive task and the computational budget assigned to the evaluation of solutions must be limited. We present a detailed study that analyses the effect of the learning conditions on the optimization strategies evolved by the Hyper-Heuristic framework. Results show that the computational budget allocation impacts the structure and quality of the learned architectures. We also present experimental results showing that the best learned strategies are competitive with state-of-the-art hand designed algorithms in unseen instances of the knapsack problem.

Towards the evolution of multi-layered neural networks: a dynamic structured grammatical evolution approach

Current grammar-based NeuroEvolution approaches have several shortcomings. On the one hand, they do not allow the generation of Artificial Neural Networks (ANNs) composed of more than one hidden-layer. On the other, there is no way to evolve networks with more than one output neuron. To properly evolve ANNs with more than one hidden-layer and multiple output nodes there is the need to know the number of neurons available in previous layers. In this paper we introduce Dynamic Structured Grammatical Evolution (DSGE): a new genotypic representation that overcomes the aforementioned limitations. By enabling the creation of dynamic rules that specify the connection possibilities of each neuron, the methodology enables the evolution of multi-layered ANNs with more than one output neuron. Results in different classification problems show that DSGE evolves effective single and multi-layered ANNs, with a varying number of output neurons.

An Evolutionary Approach to the Full Optimization of the Traveling Thief Problem

Real-World problems usually consist of several different small sub-problems interacting with each other. These interactions promote a relation of interdependence, where the quality of a solution to one sub-problem influences the quality of another partial solution. The Traveling Thief Problem (TTP) is a recent benchmark that results from the combination of the Traveling Salesman Problem (TSP) and the Knapsack Problem (KP). Thus far, existing approaches solve the TTP by fixing one of the components, usually the TSP, and then tackling the KP. We follow in a different direction and propose an Evolutionary Algorithm that addresses both sub-problems at the same time. Experimental results show that solving the TTP as whole creates conditions for discovering solutions with enhanced quality, and that fixing one of the components might compromise the overall results.

PSO-CGO: A Particle Swarm Algorithm for Cluster Geometry Optimization

In this paper the authors present PSO-CGO, a novel particle swarm algorithm for cluster geometry optimization. The proposed approach combines a steady-state strategy to update solutions with a structural distance measure that helps to maintain population diversity. Also, it adopts a novel rule to update particles, which applies velocity only to a subset of the variables and is therefore able to promote limited modifications in the structure of atomic clusters. Results are promising, as PSO-CGO is able to discover all putative global optima for short-ranged Morse clusters between 30 and 50 atoms. A comprehensive analysis is presented and reveals that the proposed components are essential to enhance the search effectiveness of the PSO.

Automatic generation of neural networks with structured Grammatical Evolution

The effectiveness of Artificial Neural Networks (ANNs) depends on a non-trivial manual crafting of their topology and parameters. Typically, practitioners resort to a time consuming methodology of trial-and-error to find and/or adjust the models to solve specific tasks. To minimise this burden one might resort to algorithms for the automatic selection of the most appropriate properties of a given ANN. A remarkable example of such methodologies is Grammar-based Genetic Programming. This work analyses and compares the use of two grammar-based methods, Grammatical Evolution (GE) and Structured Grammatical Evolution (SGE), to automatically design and configure ANNs. The evolved networks are used to tackle several classification datasets. Experimental results show that SGE is able to automatically build better models than GE, and that are competitive with the state of the art, outperforming hand-designed ANNs in all the used benchmarks.

A Comparative Study of Different Grammar-Based Genetic Programming Approaches

Grammars are useful formalisms to specify constraints, and not surprisingly, they have attracted the attention of Evolutionary Computation (EC) researchers to enforce problem restrictions. Context-Free-Grammar GP (CFG-GP) established the foundations for the application of grammars in Genetic Programming (GP), whilst Grammatical Evolution (GE) popularised the use of these approaches, becoming one of the most used GP variants. However, studies have shown that GE suffers from issues that have impact on its performance. To minimise these issues, several extensions have been proposed, which made the distinction between GE and CFG-GP less noticeable. Another direction was followed by Structured Grammatical Evolution (SGE) that maintains the separation between genotype and phenotype from GE, but overcomes most of its issues. Our goal is to perform a comparative study between CFG-GP, GE and SGE to examine their relative performance. The results show that in most of the selected benchmarks, CFG-GP and SGE have a similar performance, showing that SGE is a good alternative to GE.

Diet, Lifestyles, Family History, and Prostate Cancer Incidence in an East Algerian Patient Group.

Prostate cancer (PC) is the fourth most common cancer in men and the sixth leading cause of death in Algeria. To examine the relationship between lifestyle factors, including diet, and family history and PC risk, a case-control study was performed in an eastern Algerian population, comprising 90 patients with histologically confirmed PC and 190 controls. Data collection was carried out through a structured questionnaire and statistical analysis was performed to evaluate the different variables. The data showed that consumption of lamb and beef meat and high intake of animal fat and dairy products increased PC risk. Seven to thirteen vegetables servings per week and fourteen or more servings decreased PC risk by 62% and 96%, respectively. Seven to fourteen fruit servings per week decrease PC risk by 98%. Green tea consumption reduced the risk of PC but the results were statistically borderline. Increased risk was observed for individuals with family history of PC in first and in second degree. A positive strong association was also found for alcohol and smoking intake and a dose-response relationship existed for quantity and history of smoking. This study suggests that dietary habits, lifestyle factors, and family history have influence on the development of PC in Algerian population.