This study proposes a new strategy to improve the performance of the algorithms of the Fish School Search (FSS) family via the individualization of the step-size of each fish. We propose to be calculated in two different manners: using individual weight or using individual fitness, depending on the chosen variation of the proposed technique. Our methods were tested on the original FSS, on the Weight based Fish School Search (wFSS) and on the Multi Objective Fish School Search (MOFSS) algorithms. The benchmark functions of the Congress on Evolutionary Computation, The Genetic and Evolutionary Computation Conference (CEC'2020, CEC'2013, and GECCO'2016) and the DTLZ test suite were used to assess the experimental results, which yielded that all variants of the FSS algorithm tested have been improved in the majority of the scenarios.

For companies to have a competitive advantage, they need to extract relevant information from data and for that, they need to complement their own data with other data sources. Data marketplaces are platforms on which data providers and data consumers do business. However, every data interaction incurs a monetary cost. Therefore, data consumers are interested in buying a set of interesting views that fit their (goal and) budget. Views allow data to be represented visually, enabling users to make sense of patterns and insights. Besides allowing for easier cost control, buying a set of views bundled together increases the chance of finding what consumers want over buying them view-by-view. Selecting the suitable views to compose an interesting bundle is non-trivial, due to the vast number of view combinations that potentially meet the data consumer’s needs. In this paper, we address the problem of view bundle recommendation in data marketplaces, in which the utility of a bundle depends on the interplay among candidate views. We propose the use of Self-Organizing Maps as a means to compute this interplay and use a Genetic Algorithm to design near-optimal bundles. Our empirical results demonstrate that our approach can effectively aid data consumers to find relevant view bundles under budget constraints.

On the one hand, public expenditure can be decomposed into several levels of expenditure regarding each administrative unit and expenditure nature groups that happens thru time. Therefore, it can be considered as a hierarchical time series. On the other hand, accurate forecasting methods are desirable for planning and management due to the need to identify and anticipate future scenarios. Thus, considering the hierarchical nature of the problem, this work aims to develop a a predictive model which takes into consideration the hierarchical structure of public expenditure in order to assure financial coherence and achieve improved results. Experimentally, this work uses time series of public expenditure execution in the State of Pernambuco, Brazil. The experiments were conducted on two axes in order to establish a multilevel analysis considering conciliatory approaches of the hierarchical levels of expenditure and the predictive models. By applying these methods, valuable public expenditure forecasts in the State of Pernambuco considering different hierarchical levels were produced.

We present PALLAS, a practical method for gene regulatory network (GRN) inference from time series data, which employs penalized maximum likelihood and particle swarms for optimization. PALLAS is based on the Partially-Observable Boolean Dynamical System (POBDS) model and thus does not require ad-hoc binarization of the data. The penalty in the likelihood is a LASSO regularization term, which encourages the resulting network to be sparse. PALLAS is able to scale to networks of realistic size under no prior knowledge, by virtue of a novel continuous-discrete Fish School Search particle swarm algorithm for efficient simultaneous maximization of the penalized likelihood over the discrete space of networks and the continuous space of observational parameters. The performance of PALLAS is demonstrated by a comprehensive set of experiments using synthetic data generated from real and artificial networks, as well as real time series microarray and RNA-seq data, where it is compared to several other well-known methods for gene regulatory network inference. The results show that PALLAS can infer GRNs more accurately than other methods, while being capable of working directly on gene expression data, without need of ad-hoc binarization. PALLAS is a fully-fledged program, written in python, and available on GitHub (https://github.com/yukuntan92/PALLAS).

Many real-life engineering applications are optimization problems. To find the best configuration of variables to minimize costs and maximize efficiency are typically used engineering software as CAD, CAE and CAM. In this context, Machine Learning can be used to automate and improve this type of application. This despite, those searches not seldomly evoke unreliable areas and suggest risky solutions. Because of inaccuracies, volatility, unfeasibility, and specificities of real environments, the easy incorporation of cultural practices (i.e. normative, situational, domain and historic knowledge) and as well as the production of multiple acceptable solutions for a problem is always welcome, especially in Engineering. The present article put forward a hybridization of a multi-modal algorithm (Weight-Based Fish School Search - wFSS) with Cultural Algorithms' belief space. New cwFSS is able to guide the optimization also considering normative knowledge from experts, technical literature and problem domain readily available knowledge to prevent the incorporation of constraints into the fitness function. We also evaluated the use of temporal knowledge to guide the simulation. The proposed method was tested in a thermal power plant efficiency optimization and compared with standard wFSS and the Niching Migratory Multi-Swarm Optimizer (NMMSO), winner of CEC'2015 niching competition. As results, cwFSS has outperformed at times NMMSO about time, fitness and variability, as well as traditional wFSS about time, stability, safeness and variability of the multimodal solutions. By avoiding penalties, the appropriation of a priori search directly into the search can effectively and elegantly help better support for engineering decisions.

Motion and cognitive impairments can make the life of elderly people more difficult. Physiotherapists seek solutions to increase the quality of life of their patients through exercises and monitoring. Analysis of gait and cognitive data, collected during one year, showed patients’ responses during treatment. In this project, we used age as a feature to compare the pacient’s age and the predicted age value from gait data. The aim of this research is to show the most important features that contributed to the result of the age prediction. Support Vector Machine (SVM), Multi Layer Perceptron (MLP) and LightGBM are algorithms that had the lowest estimation errors for age prediction. Next, we carried out experiments to identify the most relevant features for the age difference. SHappley Additive exPlanations (SHAP) and Local Interpretable Model-Agnostic Explanations (LIME) were applied for global and local interpretation, respectively. The results showed that groups of features could be identified during the interpretation. These preliminary results encourage our research to apply XAI and reduce the features dimensionality, providing more confidence as well as simplicity in clinical applications that use Artificial Intelligence.

The paper investigates the suitability of a Learning Classifier System (LCS) implementation for mimicking human decision making in agent based social simulations incorporating network effects. Model behavior is studied for three distinct scenario settings. We provide proof of concept for the adequacy of LCS to tackle the task at hand. Specifically, it is found that the LCS provides the agents within the simulation model with the ability to learn and to react to environmental changes while accounting for bounded rational decision making and the presence of imperfect information, as well as network effects. Moreover, it can be shown that the LCS-agents exhibit a habit like behavioural pattern.

The design of well-performing heuristic-based algorithms is a tedious and time-consuming task. For this purpose, several approaches to automate the algorithm design have been investigated over the last years. Among them, Genetic Programming-based Hyper-Heuristics (GPHH) are able to design algorithms of varying lengths and permutations of algorithm's operators. Although the parametrization of generated algorithms is crucial for further adaptation of its behavior to the underlying problems, it has been neglected or only considered to a limited extent in many applications of GPHHs. This work proposes a way to integrate GPHH with parameter configuration. In particular, a Cartesian Genetic Programming Hyper-Heuristic (CGPHH) is developed. The proposed method is applied to the multi-level capacitated production lot-sizing problem, a complex and relevant problem in production planning. Conducted experiments evaluate both its efficacy and the quality of the generated algorithms. Finally, results are compared to the CGPHH without parameter configuration and a human-designed algorithm, indicating the ability of CGPHH-PC to generate algorithms that have the best overall-performance for different cutoff times.

Process mining has emerged as a new scientific research topic on the interface between process modeling and event data gathering.In the search for process models that best fit to reality, the process discovery approach of creating referential processes from observed behavior.However, despite these methods showing relevant results, when faced with noisy and divergent tendencies they end up producing limited results.This work proposes the application of process discovery technique, combined to cluster technique k-means, to generate new process models, considering its conformance checking measures.The proposed solution is applied to an ad hoc workflow.And as a result, the use of the clustering techniques coupled with process discovery showed significant gains in the generation of process models, unlike the standard approach.

Human decision-making involves cognitive processes of selection, evaluation, and interpretation among candidate solutions in order to solve decision problems. Nonintelligent decision support systems (DSS) lack automatic interpretations, at least in a low level scale, which can lead to undesired solutions. To tackle this limitation, hence producing enhanced decision making, a hybrid intelligent decision support approach is presented, which combines case-based reasoning cycle, semiotic concepts, and self-organizing maps. In addition, a novel sign deconstruction mechanism is introduced as foundation of the new approach and affords better interpretability and contextualization of candidate solutions without compromising efficiency and precision. The obtained results confirm that our proposed approach has the potential to be readily applicable to decision problems, particularly the ones that are of subjective nature. Moreover, the put forward approach may integrate some unlikely elements of linguistics and cognitive science which could fundamentally help the enhancement of DSS.

Ant Colony Optimization (ACO) is a well known population-based algorithm used for solving combinatorial optimization problems, such as the Traveling Salesman Problem (TSP). The parallelization of ACO becomes necessary when tackling bigger instances of the TSP due to the high number of calculations performed. Many parallel approaches have been already proposed for ACO, in particular for contemporary high-performance hardware, such as GPUs. Typically, the ants are treated in parallel, since they are largely independent. In the case of the TSP, this concerns in particular the tour construction phase. Furthermore, strategies for parallelizing the pheromone deposit and evaporation phase were proposed. The achieved overall speedup hence depends on a combination of different parallelization strategies, where the impact of each strategy depends on the characteristics of the considered application problem and the hardware used. In the present paper, we aim to compare and analyze the performance of ACO implementations using distinct parallelization strategies when solving TSP instances of different magnitudes. At first, a comparison is made between a coarse-grain and a fine-grain parallel ACO. Furthermore the impact of the parallelization of the pheromone deposit process is also analyzed. The results show that there is no overall best parallelization strategy. Also, they highlight the importance of key-points that lead to a reduction of the execution time, such as the occupancy of the GPU and the work load shared among threads.

Successful applications of heuristic-based methods are able to find high-quality solutions for complex problems in a feasible timeframe. However, they are usually tailored towards the problem instances under consideration and any changes in the underlying problem structure might require a redesign of the algorithm, which is expensive and very time-consuming. This paper presents results of an automatic algorithm-generation approach used to find good-performing optimization methods for the multi-level capacitated lot-sizing problem, a relevant and hard combinatorial problem in production planning. A new template for generating algorithms is proposed for enabling the generation of different hybridizations between genetic algorithm-components and mathematical heuristics. Several experiments are carried out to evaluate the ability of the proposed method to generate competitive algorithms for benchmark instances, under consideration of different functions set and cutoff times. Results indicate that the method is able to generate heuristic algorithms that find high-quality solutions significantly faster than the compared human-designed algorithm.

This paper proposes a mechanism of dynamic adjustment of the population size of population based metaheuristics in order to balance its efficacy and efficiency. In this approach, an external trajectory based metaheuristic (MH) is used to dynamically adjust the population size of an inner population based metaheuristic. A Particle Swarm Optmization (PSO) implemented for a Compute Unified Device Architecture platform (CUDA), called CUDA-PSO, is used as inner MH, while a sequential Simulated Annealing (SA) is used as an external one. The main objective of this paper is to evaluate the SA capabilities of finding a good balance between efficiency and efficacy during the CUDA-PSO execution and to assess its adaptability to different hardwares without any prior information about the computing platform. The results show that the new approach was able to find a good balance in most cases. Also, it was observed that this approach is able to adapt its operation to different hardwares.

The Big Data era is affording a paradigm change on decision-making approaches. More and more, companies as well as individuals are relying on data rather than on the so called "gut feeling" to make decisions. However, searching the Web for carrying out purchases is not completely satisfactory yet, given the arduousness of finding suitable quality data. This has contributed to the emergence of data marketplaces as an alternative to traditional data commerce, as they provide appropriate online environments for data offering and purchasing. Nevertheless, as the number of available datasets to purchase increases, the task of buying appropriate offers is, very often, challenging. In this sense, we propose an intelligent decision support system to help buyers in purchasing data offers based on a multiple-criteria decision analysis. Experimental results show that our approach provides an interactive way that addresses buyers' needs, allowing them to state and easily refine their preferences, without any specific order, via a series of dataset recommendations.

A well defined testing strategy is essential for any software development project. Testing efforts need to be carefully planed and executed in order to ensure effectiveness. Programming failures can represent a high risk for business. In order to mitigate such risk, companies have been increasingly investing more resources on software testing. In despite of massive investments on software testing and extensive collection of static analysis techniques and tools, there are still few conclusive explanations for what causes human programming failures on software. The hypothesis investigated in this paper is that a metric based on development teams characteristics can be more effective to predict defective source code than metrics purely focused on information about source code, alone. Aiming to assist software engineers during testing initiatives, this article presents a new approach to systematically measure capacity of development teams to handle source code complexity. The proposed metric can be effective for raising information and comparing multiple development teams, planning training initiatives and prioritising testing efforts. Experiments were carried out with the entire source code base of device drivers for Linux Operating System. Our approach was able to predict, with 80% of accuracy rate, which development teams introduced more issues from 2010 to 2014.

Football (aka Soccer) is the most popular sport in the world. The popularity of the sport leads to several stories (some perhaps anecdotal) about supporters behaviors and to the emergence of rivalries such as the famous Barcelona-Real Madrid (in Spain). Little however has been done to characterize/profile online users' behaviors as football supporters and use them as an aggregate measure to club characterization. Today, the availability of data enable us to understand at a much greater scale if rivalries exist and if there are signatures that can be used to characterize supporting behavior. In this paper we use techniques from Data Science to characterize football supporters according to their activity on Twitter and to characterize clubs according to the behavior of their supporters. We show that it is possible to: (i) rank football clubs by their popularity and fans' dislike, (ii) identify the rivalries that exist between clubs and their supporters, and (iii) find specific signatures that repeat themselves across different clubs and in different countries. The results are evaluated on a large dataset of tweets relevant to major football leagues in Brazil and in the United Kingdom.

Several micro economic models allow to evaluate consumer's behavior using a utility function that is able to measure the success of an individual's decision. Such a decision may consist of a tuple of goods an individual would like to buy and hours of work necessary to pay for them. The utility of such a decision depends not only on purchase and consumption of goods, but also on fringe benefits such as leisure, which additionally increases the utility to the individual. Utility can be used then as a collective measure for the overall evaluation of societies. In this paper, we present and compare three different agent based social simulations in which the decision finding process of consumers is performed by three algorithms from swarm intelligence and evolutionary computation. Although all algorithms appear to be suitable for the underlying problem as they are based on historical information and also contain a stochastic part which allows for modeling the uncertainty and bounded rationality, they differ greatly in terms of incorporating historical information used for finding new alternative decisions. Newly created decisions that violate underlying budget constraints may either be mapped back to the feasible region, or may be allowed to leave the valid search space. However, in order to avoid biases that would disrupt the inner rationale of each meta heuristic, such invalid decisions are not remembered in the future. Experiments indicate that the choice of such bounding strategy varies according to the choice of the optimization algorithm. Moreover, it seems that each of the techniques could excel in identifying different types of individual behavior such as risk affine, cautious and balanced.

Multi-language Source Code Management systems have been largely used to collaboratively manage software development projects. These systems represent a fundamental step in order to fully use communication enhancements by producing concrete value on the way people collaborate to produce more reliable computational systems. These systems evaluate results of analyses in order to organise and optimise source code. These analyses are strongly dependent on technologies (i.e. framework, programming language, libraries) each of them with their own characteristics and syntactic structure. To overcome such limitation, source code classification is an essential preprocessing step to identify which analyses should be evaluated. This paper introduces a new approach for generating content-based classifiers by using Evolutionary Algorithms. Experiments were performed on real world source code collected from more than 200 different open source projects. Results show us that our approach can be successfully used for creating more accurate source code classifiers. The resulting classifier is also expansible and flexible to new classification scenarios (opening perspectives for new technologies).

This work further investigates how weight based FSS (i.e. almost only use of local information) can impact in the automatic splitting the school for solution of multimodal bench mark problems. The chief modification to standard FSS, in order to produce the wFSS, was the introduction of a relationship among fish solely relying on factual already existing indications of individual success. The implementation resulted in a lighter algorithm (when compared to other FSS attempts to solve multimodal problems) and a method that produces more suitable solution candidates for optimization problems. Following a complexity analysis, that reveal a reduction from O(n4) of dFSS to O(n2) of wFSS, a thorough performance comparison with two other competing techniques was carried out using four different metrics, highly appropriate to assess multiobjective optimization. The experiments showed the upper hand of wFSS in multimodal continuous optimization adherent to the design decisions of (i) no use of global information for splitting the swarm, (ii) no heavy increase in computational costs to FSS and (iii) abidance to the original principle of FSS of not using topological information to solve the optimization problem.

We are living in an era where technology has become an essential resource for modern human welfare. Critical services like water supply, energy and transportation are controlled by computational systems. These systems must be reliable and constantly audited against software and hardware failures and malicious attacks. As a preventive approach against software vulnerabilities on critical systems, this research presents applications of computational intelligence to program analysis for vulnerability checking. This paper shows that computational intelligence techniques can successfully uncover several arithmetic and memory manipulation vulnerabilities.

Companies are often susceptible to uncertainties which can disturb the achievement of their objectives. The effect of these uncertainties can be perceived as risk that will be taken. A healthful company have to anticipate undesired events by defining a process for managing risks. Risk management processes are responsible for identifying, analyzing and evaluating risky scenarios and whether they should undergo control in order to satisfy a previously defined risk criteria. Risk specialists have to consider, at the same time, many operational aspects (decision variables) and objectives to decide which and when risk treatments have to be executed. In line with that, most companies select risks to be treated by using expertise of human specialists or simple sorting heuristics based on the believed impact. Companies have limited resources (e.g. human and financial resources) and risk treatments have costs which the selection process has to deal with. Aiming to balancing the competition between risk and resource management this paper proposes a new optimization step within the standard risk management methodology created by the International Organization for Standardization (a.k.a. ISO). To test the resulted methodology, experiments based on the Non-dominated Sorting Genetic Algorithm (more specifically NSGA-II) were performed aiming to manage risk and resources of a simulated company. Results show us that the proposed approach can deal with multiple conflicting objectives reducing the risk exposure time by selecting risks to be treated according their impact and available resources.

The use of Agent-Based Social Simulations (ABSS) using random-walk algorithms as the randomization approach in an attempt to reproduce human behaviors abound in the literature. However, almost all human actions are nonrandom and include a probabilistic component. This work investigates the use of a probabilistic distribution-based randomization method based on Levy flight as an improvement to the use of traditional random-walk algorithms in cognitive agent models. The fact that Levy flight has been shown to be a more suitable approach than random walk in mimicking natural phenomena, encouraged us to assess whether it could be a more robust randomization method in our agent-based social simulations. Experiments were carried out using Levy flight as the randomization method in our new hybrid evolutionary agent model for ABSS, modeling a realistic scenario of the spread of dengue fever. Previous papers proved that this kind of agent models, which incorporates two components namely, genetic and cultural layers, produce good simulations results. The aim here is to assess whether there is qualitative and quantitative predictive power of the model. This paper looks at the following decision components variations together with the new randomization method: (i) agents only with genetic component; (ii) agents only with cultural component; and (Hi) agents with both genetic and cultural components. The results show the beneficial impact in Levy flight that leads to better prediction in all combinations of decision components.

Many problems with wireless networks are directly related to the very means used to transport data, in this case, radio waves. In addition to mis-configured equipment lack of adaptable algorithms and wireless networks are major targets for attacks. New tools to refrain that are greatly in need. Due to the fact that it is easy to attack and not so to defend wireless networks, good candidate tools would be the ones that could profit from intelligent techniques. In this paper, we use the Danger Theory (DT) and a Bayesian classifier (using naïve Bayes) embedded in a military style multi-agent system (MAS) to create a lightweight, adaptable and dynamic detection system for wireless networks (WIDS). Experimental results show that the artificial immune aspect of the proposed system is capable of detecting unknown intrusion and to identify them automatically with considerable few false alarms and low cost for the network traffic.

Automated software testing has become a fundamental requirement for several software engineering methodologies. Software development companies very often outsource the test of their products. In such cases, the hired companies sometimes have to test softwares without any access to the source code. This type of service is called black box testing, which includes presentation of some ad-hoc input to the software followed by an assessment of the outcome. The common place for black box testing is sequential approach and slow pace of work. This ineffectiveness is due to the combinatorial explosion of software parameters and payloads. This work presents a neuro-fuzzy and multi-agent system architecture for improving black box testing tools for client-side vulnerability discovery, specifically, memory corruption flaws. Experiments show the efficiency of the proposed hybrid intelligent approach over traditional black box testing techniques.

In this paper, we propose a new method for surface reconstruction based on growing self-organizing maps (SOMs), called growing self-reconstruction maps (GSRMs). GSRM is an extension of growing neural gas (GNG) that includes the concept of triangular faces in the learning algorithm and additional conditions in order to include and remove connections, so that it can produce a triangular two-manifold mesh representation of a target object given an unstructured point cloud of its surface. The main modifications concern competitive Hebbian learning (CHL), the vertex insertion operation, and the edge removal mechanism. The method proposed is able to learn the geometry and topology of the surface represented in the point cloud and to generate meshes with different resolutions. Experimental results show that the proposed method can produce models that approximate the shape of an object, including its concave regions, boundaries, and holes, if any.

The performance of particle swarm optimization (PSO) algorithms depends strongly upon the interaction among the particles. The existing communication topologies for PSO (e.g. star, ring, wheel, pyramid, von Neumann, clan, four clusters) can be viewed as distinct means to coordinate the information flow within the swarm. Overall, each particle exerts some influence among others placed in its immediate neighborhood or even in different neighborhoods, depending on the communication schema (rules) used. The neighborhood of particles within PSO topologies is determined by the particles' indexes that usually reflect a spatial arrangement. In this paper, in addition to position information of particles, we investigate the use of adaptive density-based clustering algorithm - ADACLUS - to create neighborhoods (i.e. clusters) that are formed considering velocity information of particles. Additionally, we suggest that the new clustering rationale be used in conjunction with clan-PSO main ideas. The proposed approach was tested in a wide range of well known benchmark functions. The experimental results obtained indicate that this new approach can improve the global search ability of the PSO technique.

Deny of service (DoS) detection problem is a common and annoying network difficulty, but for IEEE 802.11 standards it becomes even more troublesome. Addressing this issue, we introduce a new approach to promptly warn the user. The detection algorithm put forward, combines second generation of artificial immune systems, danger theory and multi-agent system. For the detection system, we used the dendritic cells algorithm, modified to IEEE 802.11 environments. Experimental results carried out in controlled setups have shown that the model can easily and effectively be applied for detecting DoS in IEEE 802.11 networks.

This paper investigates a new hybrid evolutionary agent model for agent-based social simulations (ABSS), which incorporates two decision components: (i) sub-symbolic (genetic) and (ii) symbolic (cultural). These components are coherently combined to produce a more plausible agent model. Experiments were carried out using the plausible agents matrix (PAX) framework, and modeled a real dengue fever spreading scenario. They aim to analyze the qualitative and quantitative predictive power of the model. Previous work has explored the impact of structuring elements on agents' behaviors and the impacts of communication mechanisms on agents' behaviors using PAX. In this paper we investigated three types of agent models regarding to the combination of decision components: (1) agents only with genetic component; (2) agents only with cultural component; and (3) agents with both genetic and cultural components. Results show the importance of each component in the model and their synergic effects when combined.

Real world engineer problems sometimes involve high dimensional spaces. It makes them hard to compute. A common approach to tackle such challenges is to apply swarm based or evolutionary algorithms. Fish School Search (FSS) is one of such techniques that excels on difficult search problems. As FSS is a recent technique and only output results were investigated so far. This paper analyzes the influence of the FSS operators on the performance of the algorithm in six benchmark functions. We assessed the influence of each swimming operator separately. We found that the volitive mechanism is the operator that affords most of the exploration abilities during the search process. The carried out assessment has also shown that, in average, the best results are obtained only when all the FSS operators are actived. It means that all operators are fairly relevant and complementary. Moreover, we compared FSS results with some PSO variations and showed that FSS outperformed these PSO algorithms in some cases.

Intelligent Decision Support Systems (iDSS) frequently rely in analytical models to improve problem solving capabilities of decision makers. A large number of intelligent algorithms solely focus on accuracy, but this is seldom the only, or even the most important issue to be considered in decision making processes. This work investigates how to incorporate user cognitive profile preferences in the intelligent decision model. Two situations were considered in this study: (i) how to conceive appropriate models when user preferences and constraints are available and (ii) how to optimize a problem solving structure if these models are already available (i.e. can be posed by its user — the decision maker). Due to their effective application in many classification problems and their high interpretability, Decision Trees were chosen as the main inference technique and were used in four benchmark databases as our proof of concept for both: conception and optimization of intelligent models. Results suggest that the proposed approach can be useful to better bridge the gap between what the user wants and what can be provided to him, by means of intelligent algorithms. This simple, yet powerful, combination affords high levels of user satisfaction and confidence because they reduce the loss of valuable qualitative information that is readily available in the decision makers' mind. Moreover it is likely to relief the number one plague in DSSs: dismissive attitude by decision makers, leading to quite often systems dismissal.

This paper investigates the impact of agents' communication on social simulations using the PAX framework. Previous works investigated the impact of structuring elements (e.g. houses, hospitals, roads, etc) on agents' behaviors. This work extends that, by investigating the combined results of different communication schemes and structural costs. To assess the plausibility of application of these ideas on real world problems, we modeled a small Brazilian town and scenarios of dengue fever spreading over its tiny population. Results show the relevance of communication on the overall population dynamics; combined results also reveal some curious social epiphenomena.

Multispectral image analysis is a relatively promising field of research with applications in several areas, such as medical imaging and satellite monitoring. However, a considerable number of current methods of analysis are based on parametric statistics. Alternatively, some methods in computational intelligence are inspired by biology and other sciences. Here we claim that philosophy can be also considered as a source of inspiration. This work proposes the objective dialectical method (ODM), which is a computational intelligent method for classification based on the philosophy of praxis. Here, ODM is instrumental in assembling evolvable mathematical tools to analyze multispectral images. In the case study described in this paper, such multispectral images are composed of diffusion weighted (DW) magnetic resonance (MR) images. The results are compared to ground-truth images produced by polynomial networks using a morphological similarity index.

This paper presents an enhanced version of the AED (Appropriate Executive Decisions) algorithm, which is based on biological immune system (BIS) and whose purpose is the generation of appropriate executive decisions aimed at business environments. A new metric has been incorporated to the algorithm and a larger and more representative database was used to train and validate results. Moreover, this paper offers better directions on how to apply AED in executive decisions, affording the learning process quality improvement through immunoinformatics concepts, namely decision cells, thereby producing more appropriate executive decisions. Experiments were carried out with executive officers experienced in executive decisions in order to suitably validate the appropriateness of responses generated by the enhanced AED algorithm.

In order to appropriately tackle the complexity of real world problems, decision makers often use special support tools. Comprising an important class of such tools, intelligent decision support systems (iDSS) are able to not only help on the decision making process, but also improve their performance through time. Very often the use of intelligent techniques in iDSS focuses only on the reasoning mechanism. However, more than in conventional systems, a flexible interface can unleash abilities not commonly afforded to the decision maker. Flexibility here is a means to facilitate the acquisition of: (i) problem information requirements and (ii) profile of computer-user interaction. This work puts it out an interaction model based on evolutionary computation that is able to provide semi-automatic parameterization of decision trees of iDSS. As a proof of concept, experiments were conducted using four benchmark databases including several distinct features and decision scenarios. Results suggest that the proposed method is indeed useful to provide good interface adaptation (i.e. flexibility). Our approach made easier the decision task as problem information requirements and interaction profile were gathered and utilized to reframe the interface.

In this paper we propose a two-level MOEA to help on the sugarcane harvest decision support. This problem is multi-objective in nature, as it contains agronomical and logistic objectives considered simultaneously. Two different sets of heuristics were used during harvest decisions, namely crisp and fuzzy prioritization schema. They are both tested and compared here with regards to effective help to decision makers - via traditional metrics and attainment to decision scenarios. Simulations show that the productivity was increased in all hypothetical scenarios investigated because of the two-level MOEA.

Hybrid intelligent systems (HIS) are very successful in tackling problems comprising of more than one distinct computational subtask. For instance, decision-making problems are good candidates for HIS because of their frequent dual nature. This is because supporting decision-making most often involves two phases: (i) forecasting decision scenarios and (ii) searching in those scenarios. In addition to reducing the inherent uncertainty and effort in decision making, previous works in the area of decision support have shown that some of the inconveniences of the dasiaInverse Problempsila can be overcome by the use ofHybridIntelligentDecisionSuites(HIDS). This paper extends HIDS by including a third module that deals with multi-objective (MO) tasks through Evolutionary Multi-Objective Optimization (EMOO). This EMOO module helps by creating the Pareto front for each forecast scenario produced by Artificial Neural Networks (ANN), acting here as the predictive engine of the decision support system. In order to interface better with decision makers, we use a fuzzy-heuristic module of the original HIDS. To test this concept we have applied our new approach to two distinct problems: (1) diagnosis of heart diseases (of the proben-1 data-set) and (2) automobile feature selection (of UCI data-set). Results have indicated that this new ensemble of intelligent techniques enhances the quality of decision making.

Search problems are sometimes hard to compute. This is mainly due to the high dimensionality of some search spaces. Unless suitable approaches are used, search processes can be time-consuming and ineffective. Nature has evolved many complex systems able to deal with such difficulties. Fish schools, for instance, benefit greatly from the large number of constituent individuals in order to increase mutual survivability. In this paper we introduce a novel approach for searching in high-dimensional spaces taking into account behaviors drawn from fish schools. The derived algorithm - Fish-School Search (FSS) - is mainly composed of three operators: feeding, swimming and breeding. Together these operators afford the evoked computation: (i) wide-ranging search abilities, (ii) automatic capability to switch between exploration and exploitation, and (iii) self-adaptable global guidance for the search process. This paper includes a detailed description of the novel algorithm. Finally, we present simulations where the FSS algorithm is compared with, and in some cases outperforms, well-known intelligent algorithms such as Particle Swarm Optimization in high-dimensional searches.

This paper instantiate the sugarcane harvest problem as a multiple knapsack problem incorporating logistic data in its formulation. Different combinations of data representations, genetic operators and multi-objective (MO) evolutionary algorithms to solve the problem are evaluated. The proposed approach produced results that considered aspects such as output quality (i.e. relevance to decision maker), solutions spread and run-time. Tests carried out have used real data from two sugarcane mills. Finally, a MO interpretation of generated results is also suggested.

The world's demand for sugar and particularly for renewable fuels such as ethanol requires an increase in production in sugar mills. The use of artificial neural networks (ANN) posed as a predictive core associated with the algorithm NSGA-II aims at helping decision makers to optimize the multi-objective harvest problem. This paper presents two approaches and the good results achieved as compared with other classical techniques.

This work introduces a new method for surface reconstruction based on Growing Self-organizing Maps, which learn 3D coordinates of each vertex in a mesh as well as they learn the topology of the input data set. Each map grows incrementally producing meshes of different resolutions, according to the application needs. Another highlight of the presented algorithm refers to the reconstruction time, which is independent from the size of the input data. Experimental results show that the proposed method can produce models that approximate the shape of an object, including its concave regions and holes, if any.

This work presents a suite of hybrid intelligent techniques helpful in decision making, the hybrid intelligent decision suite (HIDS). The system is composed of two complementary modules, one for forecasting new decision variables and the other, for searching among generated results of candidate decisions. Using this synergistic approach, HIDS is also suitable to obtain conditioning factors leading to desired decision, thus, overcoming some of the challenges posed by the 'inverse problem'. To test this concept we have applied our approach on two distinct problems: (1) diagnosis of cardiologic diseases (of the proben-1 data-set) and (2) automobile feature selection (of UCI data-set). In the simulations carried out here, the HIDS comprised artificial neural networks (ANNs) and fuzzy logic controllers. Results proved that the ideas presented here can be effective to assemble tools which reduce uncertainty and improve quality in decision making about future scenarios.

This work presents a new architecture of artificial neural networks -Venn Networks, which produce localized activations in a 2D map while executing simple cognitive tasks. These activations resemble the ones observed in patches of the cerebral cortex when inspected by functional imaging methods such as fMRI. Venn-networks allow simultaneous incorporation of four distinct and independent concepts, all present in biological neural network. These concepts are (a) cyto-architectonic regions, (b) localization of functional activations, (c) complex pattern of intra-/interregional connectivity, and (d) definable damages to the neurons and axons. The dynamics of Venn-networks is highly influenced by these concepts. The proposed architecture incorporates both unsupervised and supervised learning paradigms; it also implements open and closed loops that can be assembled with afferent, efferent and U-flber type of connections. Venn-networks were devised to integrate in one single model the topographical representation of neural activations and also functional results evoked by these activations. Following the description of the architecture and its components, we present some simulation results that implement above-mentioned concepts (a), (b) and (c). In those simulations, virtual fingers are controlled by Venn-networks similarly to the sensorimotor feedback that controls fine movements of fingers in the CNS. The trained Venn-networks emulate the finger movements of a piano player performing The Sonata Facile of Mozart.

Novelty detection in time series is an important problem with application in different domains. such as machine failure detection, fraud detection and auditing. We have previously proposed a method for time series novelty detection based on classification of time series windows by RBF-DDA neural networks. The paper proposes a method to be used in conjunction with this time series novelty detection method whose aim is to improve performance by adequately selecting the window size and the RBF-DDA parameter values. The method was evaluated on six real-world time series and the results obtained show that it greatly improves novelty detection performance.