Pub Date : 2013-06-20DOI: 10.1109/CEC.2013.6557585
Mostafa Z. Ali, Noor H. Awad, R. Reynolds
Many evolutionary computational models have been introduced for solving engineering optimization problems that usually intend to find the global optimum solution. These methods, however, expose high computational effort and lack the diversity of the population and hence remain trapped in a local optimum. In this paper, we propose new hybrid optimization model, where a version of niche Cultural Algorithm is integrated with Tabu Search to guide the fittest individuals to new promising areas, aiming to escape local optima. The proposed approach significantly improves the performance of Cultural Algorithm by maintaining a high diversity among the population of problem solvers. This helps avoid premature and enhances located solutions. The technique is tested using a set of real-parameter optimization benchmark problems. The results in all cases indicate that the proposed method is capable of obtaining the optimal solutions with small number of function evaluations.
{"title":"Hybrid niche Cultural Algorithm for numerical global optimization","authors":"Mostafa Z. Ali, Noor H. Awad, R. Reynolds","doi":"10.1109/CEC.2013.6557585","DOIUrl":"https://doi.org/10.1109/CEC.2013.6557585","url":null,"abstract":"Many evolutionary computational models have been introduced for solving engineering optimization problems that usually intend to find the global optimum solution. These methods, however, expose high computational effort and lack the diversity of the population and hence remain trapped in a local optimum. In this paper, we propose new hybrid optimization model, where a version of niche Cultural Algorithm is integrated with Tabu Search to guide the fittest individuals to new promising areas, aiming to escape local optima. The proposed approach significantly improves the performance of Cultural Algorithm by maintaining a high diversity among the population of problem solvers. This helps avoid premature and enhances located solutions. The technique is tested using a set of real-parameter optimization benchmark problems. The results in all cases indicate that the proposed method is capable of obtaining the optimal solutions with small number of function evaluations.","PeriodicalId":211988,"journal":{"name":"2013 IEEE Congress on Evolutionary Computation","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126305036","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-06-20DOI: 10.1109/CEC.2013.6557721
Oliver Kramer, U. Schlachter, Valentin Spreckels
Constraints can make a hard optimization problem even harder. We consider the blackbox scenario of unknown fitness and constraint functions. Evolution strategies with their self-adaptive step size control fail on simple problems like the sphere with one linear constraint (tangent problem). In this paper, we introduce an adaptive penalty function oriented to Rechenberg's 1/5th success rule: if less than 1/5th of the candidate population is feasible, the penalty is increased, otherwise, it is decreased. Experimental analyses on the tangent problem demonstrate that this simple strategy leads to very successful results for the high-dimensional constrained sphere function. We accelerate the approach with two regression meta-models, one for the constraint and one for the fitness function.
{"title":"An adaptive penalty function with meta-modeling for constrained problems","authors":"Oliver Kramer, U. Schlachter, Valentin Spreckels","doi":"10.1109/CEC.2013.6557721","DOIUrl":"https://doi.org/10.1109/CEC.2013.6557721","url":null,"abstract":"Constraints can make a hard optimization problem even harder. We consider the blackbox scenario of unknown fitness and constraint functions. Evolution strategies with their self-adaptive step size control fail on simple problems like the sphere with one linear constraint (tangent problem). In this paper, we introduce an adaptive penalty function oriented to Rechenberg's 1/5th success rule: if less than 1/5th of the candidate population is feasible, the penalty is increased, otherwise, it is decreased. Experimental analyses on the tangent problem demonstrate that this simple strategy leads to very successful results for the high-dimensional constrained sphere function. We accelerate the approach with two regression meta-models, one for the constraint and one for the fitness function.","PeriodicalId":211988,"journal":{"name":"2013 IEEE Congress on Evolutionary Computation","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128174585","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-06-20DOI: 10.1109/CEC.2013.6557562
M. Couceiro, R. Rocha, N. Ferreira, P. A. Vargas
The Robotic Darwinian Particle Swarm Optimization (RDPSO) recently introduced in the literature has the ability to dynamically partition the whole population of robots based on simple “punish-reward” rules. Although this evolutionary algorithm enables the reduction of the amount of required information exchange among robots, a further analysis on the communication complexity of the RDPSO needs to be carried out so as to evaluate its scalability. This paper analyses the architecture of the RDPSO communication system, thus describing the dynamics of the communication data packet structure shared between teammates. Moreover, a set of simple communication rules is also proposed in order to reduce the communication overhead within swarms of robots. Experimental results with teams of 15 real robots show that the proposed methodology reduces the communication overhead, thus improving the scalability and applicability of the RDPSO algorithm.
{"title":"Darwinian Robotic Swarms for exploration with minimal communication","authors":"M. Couceiro, R. Rocha, N. Ferreira, P. A. Vargas","doi":"10.1109/CEC.2013.6557562","DOIUrl":"https://doi.org/10.1109/CEC.2013.6557562","url":null,"abstract":"The Robotic Darwinian Particle Swarm Optimization (RDPSO) recently introduced in the literature has the ability to dynamically partition the whole population of robots based on simple “punish-reward” rules. Although this evolutionary algorithm enables the reduction of the amount of required information exchange among robots, a further analysis on the communication complexity of the RDPSO needs to be carried out so as to evaluate its scalability. This paper analyses the architecture of the RDPSO communication system, thus describing the dynamics of the communication data packet structure shared between teammates. Moreover, a set of simple communication rules is also proposed in order to reduce the communication overhead within swarms of robots. Experimental results with teams of 15 real robots show that the proposed methodology reduces the communication overhead, thus improving the scalability and applicability of the RDPSO algorithm.","PeriodicalId":211988,"journal":{"name":"2013 IEEE Congress on Evolutionary Computation","volume":"25 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125746939","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-06-20DOI: 10.1109/CEC.2013.6557748
Barend J. Leonard, A. Engelbrecht
This paper investigates whether the optimal parameter configurations for particle swarm optimizers (PSO) change when changes in the search landscape occur. To test this, specific environmental changes that may occur during dynamic function optimization are deliberately constructed, using the moving peaks function generator. The parameters of the chargedand quantum PSO algorithms are then optimized for the initial environment, as well as for each of the constructed problems. It is shown that the optimal parameter configurations for the various environments differ not only with respect to the initial optimal configurations, but also with respect to each other. The results lead to the conclusion that PSO parameters need to be re-optimized or selfadapted whenever environmental changes are detected.
{"title":"On the optimality of particle swarm parameters in dynamic environments","authors":"Barend J. Leonard, A. Engelbrecht","doi":"10.1109/CEC.2013.6557748","DOIUrl":"https://doi.org/10.1109/CEC.2013.6557748","url":null,"abstract":"This paper investigates whether the optimal parameter configurations for particle swarm optimizers (PSO) change when changes in the search landscape occur. To test this, specific environmental changes that may occur during dynamic function optimization are deliberately constructed, using the moving peaks function generator. The parameters of the chargedand quantum PSO algorithms are then optimized for the initial environment, as well as for each of the constructed problems. It is shown that the optimal parameter configurations for the various environments differ not only with respect to the initial optimal configurations, but also with respect to each other. The results lead to the conclusion that PSO parameters need to be re-optimized or selfadapted whenever environmental changes are detected.","PeriodicalId":211988,"journal":{"name":"2013 IEEE Congress on Evolutionary Computation","volume":"373 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115785906","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-06-20DOI: 10.1109/CEC.2013.6557581
V. N. L. Izidoro, L. Castro, Angelo C. Loula
Calangos is an under development educational game based on the fauna and flora of the desert-like field of the sand dunes in the middle São Francisco River, located within the Caatinga biome in Brazil. One of the player's goals is to manage the behavior of species of lizards that inhabit this biome, with consequences to their ecology and evolution. For the development of the game a genetic-evolutionary model, embedded in a simulator, is proposed. This model will be used to simulate predator-prey dynamics based on the Evolutionary Biology and Ecology literature. The objective of this paper is to introduce the genetic-evolutionary model embedded in the simulator and present some key experimental findings. It will be shown that under certain environmental conditions lizard populations are only able to survive if allowed to evolve. The results will also show the main causes of death (malnutrition, dehydration, predation or aging), the diet preferences (vegetables or insects) of lizards and their relationship with specific environmental conditions.
Calangos是一款正在开发中的教育游戏，基于位于巴西Caatinga生物群系内的 o Francisco河中部沙丘的沙漠状区域的动植物。玩家的目标之一是管理栖息在这个生物群落中的各种蜥蜴的行为，并影响它们的生态和进化。对于游戏的开发，提出了一个嵌入在模拟器中的遗传进化模型。该模型将用于模拟基于进化生物学和生态学文献的捕食者-猎物动力学。本文的目的是介绍嵌入在模拟器中的遗传进化模型，并介绍一些关键的实验结果。它将表明，在一定的环境条件下，蜥蜴种群只有在允许进化的情况下才能生存。研究结果还将显示蜥蜴死亡的主要原因(营养不良、脱水、捕食或衰老)、饮食偏好(蔬菜或昆虫)以及它们与特定环境条件的关系。
{"title":"A genetic-evolutionary model to simulate population dynamics in the Calangos game","authors":"V. N. L. Izidoro, L. Castro, Angelo C. Loula","doi":"10.1109/CEC.2013.6557581","DOIUrl":"https://doi.org/10.1109/CEC.2013.6557581","url":null,"abstract":"Calangos is an under development educational game based on the fauna and flora of the desert-like field of the sand dunes in the middle São Francisco River, located within the Caatinga biome in Brazil. One of the player's goals is to manage the behavior of species of lizards that inhabit this biome, with consequences to their ecology and evolution. For the development of the game a genetic-evolutionary model, embedded in a simulator, is proposed. This model will be used to simulate predator-prey dynamics based on the Evolutionary Biology and Ecology literature. The objective of this paper is to introduce the genetic-evolutionary model embedded in the simulator and present some key experimental findings. It will be shown that under certain environmental conditions lizard populations are only able to survive if allowed to evolve. The results will also show the main causes of death (malnutrition, dehydration, predation or aging), the diet preferences (vegetables or insects) of lizards and their relationship with specific environmental conditions.","PeriodicalId":211988,"journal":{"name":"2013 IEEE Congress on Evolutionary Computation","volume":"585 2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115855527","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-06-20DOI: 10.1109/CEC.2013.6557894
R. L. While, Yick F. Sun, L. Barone
Area surveillance is the problem of continuously monitoring a given area for intruders or for unexpected events. Recent work has focused on the use of autonomous teams of agents for surveillance, which creates a significant planning problem. We describe an algorithm for planning in area surveillance that uses the recently-developed evolutionary optimisation technique of market-based programming, where agents develop good surveillance plans by trading tasks between them according to self-interested free-market principles. This approach is robust and scalable and it deals well with heterogeneous and dynamic environments. Experiments show that our market-based algorithm can generate good solutions to the area surveillance problem.
{"title":"A market-based approach to planning in area surveillance","authors":"R. L. While, Yick F. Sun, L. Barone","doi":"10.1109/CEC.2013.6557894","DOIUrl":"https://doi.org/10.1109/CEC.2013.6557894","url":null,"abstract":"Area surveillance is the problem of continuously monitoring a given area for intruders or for unexpected events. Recent work has focused on the use of autonomous teams of agents for surveillance, which creates a significant planning problem. We describe an algorithm for planning in area surveillance that uses the recently-developed evolutionary optimisation technique of market-based programming, where agents develop good surveillance plans by trading tasks between them according to self-interested free-market principles. This approach is robust and scalable and it deals well with heterogeneous and dynamic environments. Experiments show that our market-based algorithm can generate good solutions to the area surveillance problem.","PeriodicalId":211988,"journal":{"name":"2013 IEEE Congress on Evolutionary Computation","volume":"7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115972352","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-06-20DOI: 10.1109/CEC.2013.6557727
Eduardo Vázquez-Fernández, C. Coello
This paper employs an evolutionary algorithm to adjust the weights of the evaluation function of a chess engine. The selection mechanism of this algorithm chooses the virtual players (individuals in the population) that have the highest number of problems properly solved from a database of tactical and positional chess problems. This method has as its main advantage that we only mutate those weights involved in the solution of the current problem. Furthermore, the mutation mechanism is based on a Gaussian distribution whose standard deviation is adapted through the number of problems solved by each virtual player. We show here how, with the use of this method, we were able to increase the rating of our chess engine in 557 Elo points (from 1760 to 2317).
{"title":"An adaptive evolutionary algorithm based on tactical and positional chess problems to adjust the weights of a chess engine","authors":"Eduardo Vázquez-Fernández, C. Coello","doi":"10.1109/CEC.2013.6557727","DOIUrl":"https://doi.org/10.1109/CEC.2013.6557727","url":null,"abstract":"This paper employs an evolutionary algorithm to adjust the weights of the evaluation function of a chess engine. The selection mechanism of this algorithm chooses the virtual players (individuals in the population) that have the highest number of problems properly solved from a database of tactical and positional chess problems. This method has as its main advantage that we only mutate those weights involved in the solution of the current problem. Furthermore, the mutation mechanism is based on a Gaussian distribution whose standard deviation is adapted through the number of problems solved by each virtual player. We show here how, with the use of this method, we were able to increase the rating of our chess engine in 557 Elo points (from 1760 to 2317).","PeriodicalId":211988,"journal":{"name":"2013 IEEE Congress on Evolutionary Computation","volume":"9 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131354963","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-06-20DOI: 10.1109/CEC.2013.6557776
J. R. Bertini, M. C. Nicoletti, Liang Zhao
Graph-based representation has been successfully used to support various machine learning and data mining algorithms. The learning algorithms strongly rely on the algorithm employed for constructing the graph from input data, given as a set of vector-based patterns. A popular way to build such graphs is to treat each data pattern as a vertex; vertices are then connected according to some similarity measure, resulting in an structure known as data graph. In this paper we propose a new type of data graph, focused on data attributes, named Attribute-based Decision Graph - AbDG, suitable for supervised multiclass classification tasks. The input data for constructing an AbDG is a set of data-vectors (patterns), that can be described by either type of attributes (numeric, categorical or both). Also, algorithms for constructing such graphs and using them in classification tasks are described. An AbDG can be associated to a classifying procedure approached as a graph matching process, where the sub-graph representing a new pattern is matched against the AbDG. The proposed approach has been experimentally evaluated on classification tasks in twenty knowledge domains and the results are competitive when compared to those of two well-known classification methods (C4.5 and Multi-Interval ID3).
{"title":"Attribute-based Decision Graphs for multiclass data classification","authors":"J. R. Bertini, M. C. Nicoletti, Liang Zhao","doi":"10.1109/CEC.2013.6557776","DOIUrl":"https://doi.org/10.1109/CEC.2013.6557776","url":null,"abstract":"Graph-based representation has been successfully used to support various machine learning and data mining algorithms. The learning algorithms strongly rely on the algorithm employed for constructing the graph from input data, given as a set of vector-based patterns. A popular way to build such graphs is to treat each data pattern as a vertex; vertices are then connected according to some similarity measure, resulting in an structure known as data graph. In this paper we propose a new type of data graph, focused on data attributes, named Attribute-based Decision Graph - AbDG, suitable for supervised multiclass classification tasks. The input data for constructing an AbDG is a set of data-vectors (patterns), that can be described by either type of attributes (numeric, categorical or both). Also, algorithms for constructing such graphs and using them in classification tasks are described. An AbDG can be associated to a classifying procedure approached as a graph matching process, where the sub-graph representing a new pattern is matched against the AbDG. The proposed approach has been experimentally evaluated on classification tasks in twenty knowledge domains and the results are competitive when compared to those of two well-known classification methods (C4.5 and Multi-Interval ID3).","PeriodicalId":211988,"journal":{"name":"2013 IEEE Congress on Evolutionary Computation","volume":"14 6","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131924045","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-06-20DOI: 10.1109/CEC.2013.6557868
R. Gómez, C. Coello
The incorporation of performance indicators as the selection mechanism of a multi-objective evolutionary algorithm (MOEA) is a topic that has attracted increasing interest in the last few years. This has been mainly motivated by the fact that Pareto-based selection schemes do not perform properly when solving problems with four or more objectives. The indicator that has been most commonly used for being incorporated in the selection mechanism of a MOEA has been the hypervolume. Here, however, we explore the use of the R2 indicator, which presents some advantages with respect to the hypervolume, the main one being its low computational cost. In this paper, we propose a new MOEA called Many-Objective Metaheuristic Based on the R2 Indicator (MOMBI), which ranks individuals using a utility function. The proposed approach is compared with respect to MOEA/D (based on scalarization) and SMS-EMOA (based on hypervolume) using several benchmark problems. Our preliminary experimental results indicate that MOMBI obtains results of similar quality to those produced by SMS-EMOA, but at a much lower computational cost. Additionally, MOMBI outperforms MOEA/D in most of the test instances adopted, particularly when dealing with high-dimensional problems having complicated Pareto fronts. Thus, we believe that our proposed approach is a viable alternative for solving many-objective optimization problems.
{"title":"MOMBI: A new metaheuristic for many-objective optimization based on the R2 indicator","authors":"R. Gómez, C. Coello","doi":"10.1109/CEC.2013.6557868","DOIUrl":"https://doi.org/10.1109/CEC.2013.6557868","url":null,"abstract":"The incorporation of performance indicators as the selection mechanism of a multi-objective evolutionary algorithm (MOEA) is a topic that has attracted increasing interest in the last few years. This has been mainly motivated by the fact that Pareto-based selection schemes do not perform properly when solving problems with four or more objectives. The indicator that has been most commonly used for being incorporated in the selection mechanism of a MOEA has been the hypervolume. Here, however, we explore the use of the R2 indicator, which presents some advantages with respect to the hypervolume, the main one being its low computational cost. In this paper, we propose a new MOEA called Many-Objective Metaheuristic Based on the R2 Indicator (MOMBI), which ranks individuals using a utility function. The proposed approach is compared with respect to MOEA/D (based on scalarization) and SMS-EMOA (based on hypervolume) using several benchmark problems. Our preliminary experimental results indicate that MOMBI obtains results of similar quality to those produced by SMS-EMOA, but at a much lower computational cost. Additionally, MOMBI outperforms MOEA/D in most of the test instances adopted, particularly when dealing with high-dimensional problems having complicated Pareto fronts. Thus, we believe that our proposed approach is a viable alternative for solving many-objective optimization problems.","PeriodicalId":211988,"journal":{"name":"2013 IEEE Congress on Evolutionary Computation","volume":"39 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130075615","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-06-20DOI: 10.1109/CEC.2013.6557715
S. Naqvi, Will N. Browne, C. Hollitt
Predicting where humans look in a scene is crucial in tasks like human-computer interaction, design, graphics, image and video compression, and gaze animation. This work proposes the use of a mixed-integer constraint Genetic Algorithm (GA) for searching the optimal parameters of a bio-inspired visual saliency model for accurate prediction of human eye fixations. Bioinspired visual saliency models are complex models, mimicking the primate visual system with a vast choice of design parameters that can be tuned to achieve optimal performance. The bottom-up visual attention model used in this study was trained on three challenging image datasets from the ImgSal database using a standard performance metric (area under Receiver Operating Characteristic curve) as the fitness. To compensate for any bias of the optimized model towards the standard metric, we use two other scoring metrics to assess performance. Performance comparisons with eight state-of-the-art models have been presented for all three scoring metrics. Results show that the proposed GA optimized visual attention model provides better prediction performance than several state-of-the-art models of visual attention.
{"title":"Optimizing visual attention models for predicting human fixations using Genetic Algorithms","authors":"S. Naqvi, Will N. Browne, C. Hollitt","doi":"10.1109/CEC.2013.6557715","DOIUrl":"https://doi.org/10.1109/CEC.2013.6557715","url":null,"abstract":"Predicting where humans look in a scene is crucial in tasks like human-computer interaction, design, graphics, image and video compression, and gaze animation. This work proposes the use of a mixed-integer constraint Genetic Algorithm (GA) for searching the optimal parameters of a bio-inspired visual saliency model for accurate prediction of human eye fixations. Bioinspired visual saliency models are complex models, mimicking the primate visual system with a vast choice of design parameters that can be tuned to achieve optimal performance. The bottom-up visual attention model used in this study was trained on three challenging image datasets from the ImgSal database using a standard performance metric (area under Receiver Operating Characteristic curve) as the fitness. To compensate for any bias of the optimized model towards the standard metric, we use two other scoring metrics to assess performance. Performance comparisons with eight state-of-the-art models have been presented for all three scoring metrics. Results show that the proposed GA optimized visual attention model provides better prediction performance than several state-of-the-art models of visual attention.","PeriodicalId":211988,"journal":{"name":"2013 IEEE Congress on Evolutionary Computation","volume":"2016 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134437781","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：