Skip to main content

About this book

These transactions publish research in computer-based methods of computational collective intelligence (CCI) and their applications in a wide range of fields such as the semantic Web, social networks, and multi-agent systems. TCCI strives to cover new methodological, theoretical and practical aspects of CCI understood as the form of intelligence that emerges from the collaboration and competition of many individuals (artificial and/or natural). The application of multiple computational intelligence technologies, such as fuzzy systems, evolutionary computation, neural systems, consensus theory, etc., aims to support human and other collective intelligence and to create new forms of CCI in natural and/or artificial systems. This twenty-forth issue contains 9 carefully selected and revised contributions.p>

Table of Contents


Dynamic Topologies for Particle Swarms

The Particle Swarm Optimization (PSO) algorithm is a population-based metaheuristics in which the individuals communicate through decentralized networks. The network can be of many forms but traditionally its structure is predetermined and remains fixed during the search. This paper investigates an alternative approach. The particles are positioned on a 2-dimensional grid of nodes. During the run, they move through the network according to simple rules, while interacting with each other using signs that they leave on the nodes. The links between the particles – and consequently the information flow – are then defined at each time step by the position of the particle on the grid. As a result, each particle’s set of neighbors and connectivity degree varies during the search progress. The particles can move randomly or instead track signs left by other particles on the grid. In this paper, after a formal description of the general model, two different strategies (random and sign-based) are tested and compared to standard topologies on unimodal and multimodal functions, including a rotated and a shifted function with noise from the CEC benchmark. The experiments demonstrate that the dynamics provided by the proposed structure results in a more consistent and stable performance throughout the test set. The working mechanisms of the model are simple and easy to implement.
Carlos M. Fernandes, J. L. J. Laredo, J. J. Merelo, C. Cotta, A. C. Rosa

Evaluative Study of PSO/Snake Hybrid Algorithm and Gradient Path Labeling for Calculating Solar Differential Rotation

PSO/Snake hybrid algorithm is a merge of particle swarm optimization (PSO), a successful population based optimization technique, and the Snake model, a specialized image processing algorithm. In the PSO/Snake hybrid algorithm each particle in the population represents only a portion of the solution and the population, as a whole, will converge to the final complete solution. In this model there is a one-to-one relation between Snake model snaxels and PSO particles with the PSO’s kinematics being modified accordingly to the snake model dynamics. This paper provides an evaluative study on the performance of the customized PSO/Snake algorithm in solving a real-world problem from astrophysics domain and comparing the results with Gradient Path Labeling (GPL) image segmentation algorithm. The GPL algorithm segments the image into regions according to its intensity from where the relevant ones can be selected based on their features. A specific type of solar features called coronal bright points have been tracked in a series of solar images using both algorithms and the solar differential rotation is calculated accordingly. The final results are compared with those already reported in the literature.
Ehsan Shahamatnia, André Mora, Ivan Dorotovič, Rita A. Ribeiro, José M. Fonseca

The Uncertainty Quandary: A Study in the Context of the Evolutionary Optimization in Games and Other Uncertain Environments

In many optimization processes, the fitness or the considered measure of goodness for the candidate solutions presents uncertainty, that is, it yields different values when repeatedly measured, due to the nature of the evaluation process or the solution itself. This happens quite often in the context of computational intelligence in games, when either bots behave stochastically, or the target game possesses intrinsic random elements, but it shows up also in other problems as long as there is some random component. Thus, it is important to examine the statistical behavior of repeated measurements of performance and, more specifically, the statistical distribution that better fits them. This work analyzes four different problems related to computational intelligence in videogames, where Evolutionary Computation methods have been applied, and the evaluation of each individual is performed by playing the game, and compare them to other problem, neural network optimization, where performance is also a statistical variable. In order to find possible patterns in the statistical behavior of the variables, we track the main features of its distributions, skewness and kurtosis. Contrary to the usual assumption in this kind of problems, we prove that, in general, the values of two features imply that fitness values do not follow a normal distribution; they do present a certain common behavior that changes as evolution proceeds, getting in some cases closer to the standard distribution and in others drifting apart from it. A clear behavior in this case cannot be concluded, other than the fact that the statistical distribution that fitness variables follow is affected by selection in different directions, that parameters vary in a single generation across them, and that, in general, this kind of behavior will have to be taken into account to adequately address uncertainty in fitness in evolutionary algorithms.
Juan J. Merelo, Federico Liberatore, Antonio Fernández Ares, Rubén García, Zeineb Chelly, Carlos Cotta, Nuria Rico, Antonio M. Mora, Pablo García-Sánchez, Alberto Tonda, Paloma de las Cuevas, Pedro A. Castillo

Hybrid Single Node Genetic Programming for Symbolic Regression

This paper presents a first step of our research on designing an effective and efficient GP-based method for symbolic regression. First, we propose three extensions of the standard Single Node GP, namely (1) a selection strategy for choosing nodes to be mutated based on depth and performance of the nodes, (2) operators for placing a compact version of the best-performing graph to the beginning and to the end of the population, respectively, and (3) a local search strategy with multiple mutations applied in each iteration. All the proposed modifications have been experimentally evaluated on five symbolic regression benchmarks and compared with standard GP and SNGP. The achieved results are promising showing the potential of the proposed modifications to improve the performance of the SNGP algorithm. We then propose two variants of hybrid SNGP utilizing a linear regression technique, LASSO, to improve its performance. The proposed algorithms have been compared to the state-of-the-art symbolic regression methods that also make use of the linear regression techniques on four real-world benchmarks. The results show the hybrid SNGP algorithms are at least competitive with or better than the compared methods.
Jiří Kubalík, Eduard Alibekov, Jan Žegklitz, Robert Babuška

L2 Designer

A Tool for Genetic L-system Programming in Context of Generative Art
We propose a new format to define parametric L-systems (L2 language) and its implementation in JavaScript (L2 Designer). Our language allows us to create formal definition of the hierarchy of L-systems. The L2 Designer enables us to discover L-system grammars by means of interactive evolution - the common method used in Evolutionary art.
We provide an example of L2 program and we illustrate possibilities of L2 Designer on the two case studies. First case study was inspired by an artistic decorative floral pattern. Second case study describes the detailed process of developing a new L-system grammar that leads to the original graphics.
Tomáš Konrády, Kamila Štekerová, Barbora Tesařová

Manifold Learning Approach Toward Constructing State Representation for Robot Motion Generation

This paper presents a bottom-up approach to building internal representation of an autonomous robot. The robot creates its state space for planning and generating actions adaptively based on collected information of image features without pre-programmed physical model of the world. For this purpose, image-feature-based state space construction method is proposed using manifold learning approach. The visual feature is extracted from sample images by SIFT (scale invariant feature transform). SOM (Self Organizing Map) is introduced to find appropriate labels of image features throughout images with different configurations of robot. The vector of visual feature points mapped to low dimensional space express relation between the robot and its environment with LLE (locally linear embedding). The proposed method was evaluated by experiment with a humanoid robot collision classification and motion generation in an obstacle avoidance task.
Yuichi Kobayashi, Ryosuke Matsui

The Existence of Two Variant Processes in Human Declarative Memory: Evidence Using Machine Learning Classification Techniques in Retrieval Tasks

This work use supervised machine learning methods on fMRI brain scans, taken/measured during a memory-retrieval task, to support establishing the existence of two distinct systems for human declarative memory (“Explicit Encoding” (EE) and “Fast Mapping” (FM)). The importance of using retrieval is that it allows a direct comparison between exemplars designed to use EE and those designed to use FM. This is not directly available under acquisition tasks because of the nature of the purported memory systems since the tasks are necessarily somewhat distinct between the two systems under acquisition. This means that there could be a confounding of the distinction in the task with the difference in the representation and mechanism of the internal memory system during analysis. Retrieval tasks, on the other hand allow for identity of task. Thus this work fills a lacuna in earlier work which used memory acquisition tasks. In addition, since the data used in this work was gathered over a two day period, the classification methods is also able to identify a distinction in the consolidation of the memories in the two systems. The results presented here clearly support the existence of the two distinct memory systems.
Alex Frid, Hananel Hazan, Ester Koilis, Larry M. Manevitz, Maayan Merhav, Gal Star

Divide and Conquer Ensemble Method for Time Series Forecasting

Time series forecasting have attracted a great deal of attention from various research communities. There are many methods which divide time series into subseries. Information granules, fuzzy clustering and data segmentation are among the most popular methods in this field. However all these methods are designed to recognize dependencies between adjacent points. In order to do so, they divide the time series into time intervals. This imply some limitations in findings strongly non-local dependencies between points scatter across whole time series. The Divide and Conquer ensemble algorithm here presented was designed to address such limitations. The model samples the series into many subseries, searches for possible patterns and finally chooses the most significant subseries for further investigation. Since the prediction error evaluated on the subseries is lower than the one evaluated on the original time-series, the proposed strategy can significantly mitigate the overall prediction error. In order to evaluate the efficiency of our approach we performed the analysis on various artificial datasets. In a real world example our algorithm showed a 3-fold improvement of the accuracy with respect to other state-of-the-art methods. Although the algorithm was designed for time-series forecasting, it can be easily used for noise filtering purposes. Simulations reported in the present work illustrate the potential of the method in this field of application.
Jan Kostrzewa, Giovanni Mazzocco, Dariusz Plewczynski

Application Areas of Ephemeral Computing: A Survey

It is increasingly common that computational devices with significant computing power are underexploited. Some of the reasons for that are due to frequent idle-time or to the low computational demand of the tasks they perform, either sporadically or in their regular duty. The exploitation of this (otherwise-wasted) computational power is a cost-effective solution for solving complex computational tasks. Individually (device-wise), this computational power can sometimes comprise a stable, long-lasting availability window but it will more frequently take the form of brief, ephemeral bursts. Then, in this context a highly dynamic and volatile computational landscape emerges from the collective contribution of such numerous devices. Algorithms consciously running on this kind of environment require specific properties in terms of flexibility, plasticity and robustness. Bioinspired algorithms are particularly well suited to this endeavor, thanks to some of the features they inherit from their biological sources of inspiration, namely decentralized functioning, intrinsic parallelism, resilience, and adaptiveness. Deploying bioinspired techniques on this scenario, and conducting analysis and modelling of the underlying Ephemeral Computing environment will also pave the way for the application of other non-bioinspired techniques on this computational domain. Computational creativity and content generation in video games are applications areas of the foremost economical interest and are well suited to Ephemeral Computing due to their intrinsic ephemeral nature and the widespread abundance of gaming applications in all kinds of devices. In this paper, we will explain why and how they can be adapted to this new environment.
Carlos Cotta, Antonio J. Fernández-Leiva, Francisco Fernández de Vega, Francisco Chávez, Juan J. Merelo, Pedro A. Castillo, David Camacho, María D. R-Moreno


Additional information

Premium Partner

    Image Credits