Dynamic multi-objective optimisation problems have more than one objective, at least two objectives that are in conflict with one another and at least one objective that changes over time. These kinds of problems do not have a single optimum due to the conflict between the objectives. Therefore, a new approach is required to determine the quality of a solution. Traditionally in multi-objective optimisation (MOO) Pareto-dominance have been used to compare the quality of two solutions. However, in order to increase the speed of convergence and the diversity of the found solutions, \(\epsilon \)-dominance has been proposed. This study investigates the effect of using \(\epsilon \)-dominance for two aspects of the dynamic vector evaluated particle swarm optimisation (DVEPSO) algorithm, namely: updating the global best and managing the archive solutions. The results indicate that applying \(\epsilon \)-dominance instead of Pareto-dominance to either both of these aspects of the algorithm, or only to the global best update, does improve the performance of DVEPSO.
This paper presents the design and synthesis of a nature-inspired micro-strip patch antenna based on the sunflower structure. The antenna structure was based on the Fibonacci pattern found in a sunflower, with the antenna elements in the position of the seeds. Simulation is done using computer simulation technology Microwave studio and the geometry offers impedance bandwidth of 5.28 GHz with enhanced radiation parameters. The geometry has simple structure, therefore can be used for satellite communication applications.
The analytical observation of nature induces inspiration to propose new computational paradigms to create algorithms that solve optimization and artificial intelligence problems. The artificial vision allows establishing a problem with intelligent techniques from living systems. The bioinspired systems are presented as a set of models that are based on the behavior and the way of acting of some biological systems. These models can be expressed in data mining and operations research where the clustering is a recurrent technique in the P-median problem and territorial design. On this point, we have solved clustering problems using partitioning with bioinspired aspects and variable neighborhood search to approximate optimal solutions. In this work we have improved the search strategy: we present a bioinspired partitioning algorithm with optimization by tabu search (TS). This clustering problem under a bioinspired connotation has been proposed after observing some characteristics in common between clustering and human behavior in conflict situations, where some characteristics have been modeled.
One of the major issues that should be addressed when solving dynamic problems, is a loss of diversity. In addition, when solving multi-objective optimisation problems, one of the goals is to find a diverse set of solutions. Therefore, a key component of a dynamic multi-objective optimisation algorithm (DMOA) is an approach to increase diversity of either the dynamic multi-objective optimisation algorithm DMOA’s individuals or the guides that guide the search of the DMOA. This study investigates whether using the headless chicken macromutation operator for local guide selection improves the performance of the dynamic vector evaluated particle swarm optimisation (DVEPSO) algorithm. Results indicate that the operator does improve the accuracy of the set of solutions that is found by DVEPSO. However, fewer solutions are found.
Evolutionary Robotics (ER) makes use of evolutionary algorithms to evolve controllers and morphologies of robots. Despite successful demonstrations in laboratory experiments, ER has not been widely adopted by industry as means of robot design. A possible reason for this is that current ER approaches ignore issues that are important when designing robots for practical use. For example, the availability and cost of components used for robot construction should be considered. A robot designed by the ER process may require specialised custom components to be built to support the physical functioning of the design, if the components selected by an ER process are not widely available. Alternatively, the ER designed robot may be too expensive to be constructed. This paper demonstrates that standard off-the-shelf components can be used by the ER process to design a robot. A robot arm is used as a sample problem, which is successfully optimised to use components from a fixed list while minimising cost.
The design of an efficient and automated model for plant recognition and classification will give the possibility to people with little or no botanical knowledge to conduct field work. In this paper a feature for leaf shape analysis is presented: the Sinuosity coefficients. This new feature is based on the sinuosity measures, which is a value expressing the degree of meandering of a curve. The sinuosity coefficients is a vector of sinuosity measures characterising a given shape. The proposed shape feature is translation and scale invariant. This feature achieved a classification rate of 92% with the Multilayer Perceptron (MLP) classifier, a rate of 91% with the K-Nearest Neighbour (KNN) and a rate of 94% with the Naive Bayes classifier, on a set of leaf images from the FLAVIA dataset.
A recent direction of hyper-heuristics is the automated design of intelligent systems with the aim of reducing the man hours needed to implement such systems. One of the design decisions that often has to be made when developing intelligent systems is the low-level construction heuristic to use. These are usually rules of thumb derived based on human intuition. Generally a heuristic is derived for a particular domain. However, according to the no free lunch theorem different low-level heuristics will be effective for different problem instances. Deriving low-level heuristics for problem instances will be time consuming and hence we examine the automatic induction of low-level heuristics using hyper-heuristics. We investigate this for classical artificial intelligence. At the inception of the field of artificial intelligence search methods to solve problems were generally uninformed, such as the depth first and breadth first searches, and did not take any domain specific knowledge into consideration. As the field matured domain specific knowledge in the form of heuristics were used to guide the search, thereby reducing the search space. Search methods using heuristics to guide the search became known as informed searches, such as the best-first search, hill-climbing and the A* algorithm. Heuristics used by these searches are problem specific rules of thumb created by humans. This study investigates the use of a generative hyper-heuristic to derive these heuristics. The hyper-heuristic employs genetic programming to evolve the heuristics. The approach was tested on two classical artificial intelligence problems, namely, the 8-puzzle problem and Towers of Hanoi. The genetic programming system was able to evolve heuristics that produced solutions for 20 8-puzzle problems and 5 instances of Towers of Hanoi. Furthermore, the heuristics induced were able to produce solutions to the instances of the 8-puzzle problem which could not be solved using the A* algorithm with the number of tiles out of place heuristic and at least one admissible heuristic was evolved for all 25 problems.
Profile matching in social networks is a challenging problem. There are many previous approaches but all of them rely on data, for which there is no ground truth. Profile information is provided by the users. Therefore, there is no certainty about the reliability of this information. For this reason, we suggest using timestamps, which are generated by the social network, and device-generated geo-tags. We take a closer look at various approaches for implementation of this profile matching algorithm. In addition, possibilities for evaluation of this algorithm are outlined afterwards.
The phenomenon of social interactions is prevailing charismatically like a spider net in nowadays society despite the people busy lives. In this fashion, people willingly supply their private or public data without sensing the threat of any information theft. These kinds of information could be easily misused and could be analyzed by any third party for malicious or non-malicious purposes. In this paper, detection of irregular or anomalous individual are focused. Individual with behavioral dissimilarity are discovered and validated with the real denounced victims. An affluent feature set of 15 characteristics is anticipated for deviation detection. The kth nearest neighbour technique is applied on the Enron dataset for finding accused email users. Noteworthy outputs are achieved by implication of the KNN method.
Ensemble learning is one of the machine learning approaches, which can be described as the process of combining diverse models to solve a particular computational intelligence problem. We can find the analogy to this approach in human behavior (e.g. consulting more experts before taking an important decision). Ensemble learning is advantageously used for improving the performance of classification or prediction models. The whole process strongly depends on the process of determining the weights of base methods. In this paper we investigate different weighting schemes of predictive base models including biologically inspired genetic algorithm (GA) and particle swarm optimization (PSO) in the domain of electricity consumption. We were particularly interested in their ability to improve the performance of ensemble learning in the presence of different types of concept drift that naturally occur in electricity load measurements. The PSO proves to have the best ability to adapt to the sudden changes.
The development of sensor hardware have made it possible to transmit real time multimedia data over a wireless medium using tiny resource constrained sensors. However, in current wireless sensor networks, multimedia traffic which has stringent bandwidth and delay requirements is not distinctively differentiated from other data types during transmission which makes it difficult to meet its service requirements. Next generation wireless sensor networks are predicted to deploy a different model where service is allocated depending on the nature of data to be transmitted. Applying traditional wireless sensor routing algorithms to wireless multimedia sensor networks may lead to high delay and poor visual quality for multimedia applications. In this paper, we propose a priority based rate routing protocol that assigns priorities to traffic depending on their service requirements. We study, the performance of our proposed routing algorithm for real time traffic when mixed with three non real time traffic but with different priorities: high, medium and low priority. Initial results from the simulation show that the proposed algorithm performs better compared to two existing algorithms PCCP and CCF in terms of delay, loss and throughput.
Social dilemma is a challenge to many scientists. The Prisoner's Dilemma and Snowdrift game were the most used social dilemma models in the cooperation evolution. A particularly effect to the evolutionary process comes from population structure. By comparing population structures that amplify selection with other population structures, both analytically and numerically, we show that evolution also affected by the cost to benefit ratio and neighbor number.
Hyper-heuristics seek solution methods instead of solutions and thus provides a higher level of generality compared to bespoke metaheuristics and traditional heuristic approaches. In this paper, a hyper-heuristic is proposed to solve the ski-lodge problem which involves allocating shared-time apartments to customers during a skiing season in a way that achieves a certain objective while respecting the constraints of the problem. Prior approaches to the problem include simulated annealing and genetic algorithm. To the best of our knowledge, this is the first time the ski-lodge problem is approached from a hyper-heuristic perspective. Although the aim of hyper-heuristics is to provide good results over problem sets rather than producing best results for certain problem instances, for completeness and to get an idea of the quality of solutions, the results of the proposed hyper-heuristic are compared to that of genetic algorithm and simulated annealing. The hyper-heuristic was found to perform better than simulated annealing and comparatively to the genetic algorithm, producing better results for some of the instances. Furthermore, the hyper-heuristic has better overall performance over the problem set being considered.
Cars, trucks, and other vehicles typically run on gasoline or diesel, fuels release harmful chemicals in the air. These emissions can create a multitude of problems, including health issues and environmental degradation due to pollution. To reduce these emissions and the problems that they create, it is really important to monitor and control them more frequently. This paper tries to address these issues proposing a real-time vehicle emission monitoring and location tracking framework. The proposed framework uses two main technologies Vehicle On-Board Diagnostic (OBD-II) to monitor vehicle emission information and Assisted Global Positioning System (A-GPS) to get location of the vehicle at real-time. The existing wireless network infrastructure is used in supporting the emission and location data collection on a central server and further processing and presentation of the information is done. To evaluate the operational effectiveness of the developed framework, simulation based experiment is conducted. Real-time vehicle flow on selected roads is conducted using the microscopic simulation SUMO. Within 99 time interval integrated gas emission and location information about individual vehicle at different points of traffic roads is produced. The collected data is preprocessed, categorized and the information is proposed to be displayed on road users' mobile phone as well as on computer connected to the Internet.
Automatic programming is a concept which until today has not been fully achieved using evolutionary algorithms. Despite much research in this field, a lot of the concepts remain unexplored. The current study is part of ongoing research aimed at using evolutionary algorithms for automatic programming. The performance of two evolutionary algorithms, namely, genetic programming and grammatical evolution are compared for automatic object-oriented programming. Genetic programming is an evolutionary algorithm which searches a program space for a solution program. A program generated by genetic programming is executed to yield a solution to the problem at hand. Grammatical evolution is a variation of genetic programming which adopts a genotype–phenotype distinction and uses grammars to map from a genotypic space to a phenotypic (program) space. The study implements and tests the abilities of these approaches as well as a further variation of genetic programming, namely, object-oriented genetic programming, for automatic object-oriented programming. The application domain used to evaluate these approaches is the generation of abstract data types, specifically the class for the list data structure. The study also compares the performance of the algorithms when human programmer problem domain knowledge is incorporated and when such knowledge is not incorporated. The results show that grammatical evolution performs better than genetic programming and object-oriented genetic programming, with object-oriented genetic programming outperforming genetic programming. Future work will focus on evolution of programs that use the evolved classes.
The research presented in this paper forms part of a larger initiative aimed at creating a general game player for two player zero sum board games. In previous work, we have presented a novel heuristic based genetic programming approach for evolving game playing for the board game Othello. This study extends this work by firstly evaluating it on a different board game, namely, checkers. Secondly, the study investigates incorporating reinforcement learning to further improve evolved game playing strategies. Genetic programming evolves game playing strategies composed of heuristics, which are used to decide which move to make next. Each strategy represents a player. A separate genetic programming run is performed for each move of the game. Reinforcement learning is applied to the population at the end of a run to further improve the evolved strategies. The evolved players were found to outperform random players at checkers. Furthermore, players induced combining genetic programming and reinforcement learning outperformed the genetic programming players. Future research will look at further application of this approach to similar non-trivial board games such as chess.
The rearrangement of genomes is an important tool for studying the evolution of genomes and specifically for the construction of phylogenies. A translocation splits and combines the strings of genes of a pair of chromosomes inside a genome and is considered a suitable operation for rearrangement of genomes with multiple chromosomes. The translocation distance between two genomes is the minimum number of translocations necessary to convert one of them into the other. Computing the translocation distance between two unsigned genomes, that is the case in which the direction of the genes between the chromosomes is not considered, is known to be an MV-hard optimization problem. Among several approximation algorithms that were proposed for solving this problem, the authors introduced in a previous work a genetic algorithm approach improved with opposition based learning and memetic mechanisms. In this paper, two parallel treatments of the sequential memetic approach are introduced for solving the translocation distance problem for unsigned genomes. The first approach, computes in parallel the fitness over all individuals of a population. This method intends speeding-up the sequential memetic algorithm. The second approach, processes in parallel multiple populations and was proposed for improving precision providing solutions with a less number of translocations than the sequential memetic algorithm. Several experiments were performed with randomly generated synthetic and biologically based genomes. Results show that the parallel approaches outperform the sequential memetic algorithm.
One of the challenging issues in bioinformatics field is that, microarray datasets are imbalance in nature i.e., the majority class dominates the minority class making it difficult for the conventional classifiers to achieve accurate and useful predictions. However, some studies have addressed this issue merely by focusing on binary–class problems. In this article, an ensemble framework is proposed for multiclass imbalance classification problem that combines a meta learning algorithm ‘decorate’ with a sampling technique to deal with the problem in microarray datasets. The meta-learning algorithm builds diverse ensembles of classifiers constructing artificial samples and the sampling technique introduces bias to achieve uniform class distribution to reduce misclassification error. Experimental results on the two highly imbalanced multiclass microarray cancer datasets indicate that the technique applied provides significant improvement in comparison to other conventional ensembles.
This paper presents how design science research can be used to design and evaluate real-time road traffic state estimation framework. An integrated framework of the six process steps of design science research process model and the Matching Analysis-Projection-Synthesis (MAPS) tool was used as a research design to develop the proposed state estimation framework. The utility and efficiency of the framework was evaluated based on the adapted design science research evaluation guideline through simulation and the estimation accuracy indicated that reliable road traffic state estimation can be generated based on the developed framework.
k-means and k-medoids have been the most popular clustering algorithms based on partitioning for many decades. When using heuristics such as Lloyd’s algorithm, k-means is easy to implement and can be applied on large data sets. However, it presents drawbacks like the inefficiency of the used metric, the difficulty of the choice of the input k and the premature convergence. In contrast, k-medoids takes more time to come up with a clustering but ensures a better quality of the result. Moreover, it is more robust to noise and outliers. In this article, we design a hybrid algorithm, namely k-MM to take advantage of both algorithms. We experimented k-MM and we show that, when compared to k-means and k-medoids, it is very efficient and effective. We present also an application to image clustering and show that k-MM has the ability to discover clusters faster and more effectively than a recent work of the literature.
