Pub Date : 2016-12-01DOI: 10.1109/SSCI.2016.7849989
F. Leichsenring, W. Graf, M. Kaliske
In engineering related tasks, multiple types of neural networks are common methods of solution. Beside the different kinds of artificial neural networks, spiking neural networks (SNN) represent a continuative development in information processing within the computational units of a net. The properties of this neural type is utilized in this contribution in order to evaluate a uniaxial tension test of carbon reinforced specimen regarding the appearance of cracks in the composite structure during the experiment. The crack detection is considered as showcase for further development of evaluation methods based on SNNs with the focal point to engineering related experiments. This contribution is divided into five main parts, whereas the initial brief introduction is devoted to give an overview of neural networks and their computational units, particularly with regard to the classification of spiking neural networks. Since the proposed application of SNNs targets the evaluation of experimental data - especially crack detection - the uniaxial tension test of carbon reinforced concrete specimen is introduced, which is the basis for the experimental data. The utilized spike response model (SRM) is further presented in order to conclusively apply the method to experimental data for the purpose of crack occurrence detection within the data.
{"title":"Spiking response model for uniaxial carbon concrete experimental data","authors":"F. Leichsenring, W. Graf, M. Kaliske","doi":"10.1109/SSCI.2016.7849989","DOIUrl":"https://doi.org/10.1109/SSCI.2016.7849989","url":null,"abstract":"In engineering related tasks, multiple types of neural networks are common methods of solution. Beside the different kinds of artificial neural networks, spiking neural networks (SNN) represent a continuative development in information processing within the computational units of a net. The properties of this neural type is utilized in this contribution in order to evaluate a uniaxial tension test of carbon reinforced specimen regarding the appearance of cracks in the composite structure during the experiment. The crack detection is considered as showcase for further development of evaluation methods based on SNNs with the focal point to engineering related experiments. This contribution is divided into five main parts, whereas the initial brief introduction is devoted to give an overview of neural networks and their computational units, particularly with regard to the classification of spiking neural networks. Since the proposed application of SNNs targets the evaluation of experimental data - especially crack detection - the uniaxial tension test of carbon reinforced concrete specimen is introduced, which is the basis for the experimental data. The utilized spike response model (SRM) is further presented in order to conclusively apply the method to experimental data for the purpose of crack occurrence detection within the data.","PeriodicalId":120288,"journal":{"name":"2016 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2016-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125808989","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 : 2016-12-01DOI: 10.1109/SSCI.2016.7850107
Daniel Hennes, D. Izzo, D. Landau
We consider the problem of optimally transferring a spacecraft from a starting to a target asteroid. We introduce novel approximations for important quantities characterizing the optimal transfer in case of short transfer times (asteroid hops). We propose and study in detail approximations for the phasing value φ, for the maximum initial mass m* and for the arrival mass mf. The new approximations require orders of magnitude less computational effort with respect to state-of-the-art algorithms able to compute their ground-truth value. The accuracy of the introduced approximations is also found to be orders of magnitude superior with respect to other, commonly used, approximations based, for example, on Lambert models. Our results are obtained modelling the physics of the problem as well as employing computational intelligence techniques including the multi-objective evolutionary algorithm by decomposition framework, the hypervolume indicator and state of the art machine learning regressors.
{"title":"Fast approximators for optimal low-thrust hops between main belt asteroids","authors":"Daniel Hennes, D. Izzo, D. Landau","doi":"10.1109/SSCI.2016.7850107","DOIUrl":"https://doi.org/10.1109/SSCI.2016.7850107","url":null,"abstract":"We consider the problem of optimally transferring a spacecraft from a starting to a target asteroid. We introduce novel approximations for important quantities characterizing the optimal transfer in case of short transfer times (asteroid hops). We propose and study in detail approximations for the phasing value φ, for the maximum initial mass m* and for the arrival mass mf. The new approximations require orders of magnitude less computational effort with respect to state-of-the-art algorithms able to compute their ground-truth value. The accuracy of the introduced approximations is also found to be orders of magnitude superior with respect to other, commonly used, approximations based, for example, on Lambert models. Our results are obtained modelling the physics of the problem as well as employing computational intelligence techniques including the multi-objective evolutionary algorithm by decomposition framework, the hypervolume indicator and state of the art machine learning regressors.","PeriodicalId":120288,"journal":{"name":"2016 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2016-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115254423","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 : 2016-12-01DOI: 10.1109/SSCI.2016.7849956
Zhun Fan, Jiewei Lu, Yibiao Rong
This paper proposes a novel and simple unsupervised vessel segmentation algorithm using fundus images. At first, the green channel of a fundus image is preprocessed to extract a binary image after the isotropic undecimated wavelet transform, and another binary image from the morphologically reconstructed image. Secondly, two initial vessel images are extracted according to the vessel region features for the connected regions in binary images. Next, the regions common to both initial vessel images are extracted as the major vessels. Then all remaining pixels in two initial vessel images are processed with skeleton extraction and simple linear iterative clustering. Finally the major vessels are combined with the processed vessel pixels. The proposed algorithm outperforms its competitors when compared with other widely used unsupervised and supervised methods, which achieves a vessel segmentation accuracy of 95.8% and 95.8% in an average time of 9.7s and 14.6s on images from two public datasets DRIVE and STARE, respectively.
{"title":"Automated blood vessel segmentation of fundus images using region features of vessels","authors":"Zhun Fan, Jiewei Lu, Yibiao Rong","doi":"10.1109/SSCI.2016.7849956","DOIUrl":"https://doi.org/10.1109/SSCI.2016.7849956","url":null,"abstract":"This paper proposes a novel and simple unsupervised vessel segmentation algorithm using fundus images. At first, the green channel of a fundus image is preprocessed to extract a binary image after the isotropic undecimated wavelet transform, and another binary image from the morphologically reconstructed image. Secondly, two initial vessel images are extracted according to the vessel region features for the connected regions in binary images. Next, the regions common to both initial vessel images are extracted as the major vessels. Then all remaining pixels in two initial vessel images are processed with skeleton extraction and simple linear iterative clustering. Finally the major vessels are combined with the processed vessel pixels. The proposed algorithm outperforms its competitors when compared with other widely used unsupervised and supervised methods, which achieves a vessel segmentation accuracy of 95.8% and 95.8% in an average time of 9.7s and 14.6s on images from two public datasets DRIVE and STARE, respectively.","PeriodicalId":120288,"journal":{"name":"2016 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2016-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115553109","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 : 2016-12-01DOI: 10.1109/SSCI.2016.7849941
Q. Huynh, H. Singh, T. Ray
The process of identifying analytical relationships among variables and responses in observed data is commonly referred to as Symbolic Regression (SR). Genetic Programming is one of the commonly used approaches for SR, which operates by evolving expressions. Such relationships could be explicit or implicit in nature, of which the former has been more extensively studied in literature. Even though extensive studies have been done in SR, the fundamental challenges such as bloat, loss of diversity and accurate determination of coefficients still persist. Recently, semantics and multi-objective formulation have been suggested as potential tools to alleviate these issues by building more intelligence in the search process. However, studies along both these directions have been in isolation and applied only to selected components of SR so far. In this paper, we intend to build a framework that integrates semantics deeper into more components of SR. The framework could be operated in conventional single objective as well as multi-objective mode and is capable of dealing with both explicit and implicit functions. Semantics are used in the proposed framework for improving compactness and diversity of expressions, crossover and local exploitation. Numerical experiments are presented on a set of benchmark problems to demonstrate the strengths of the proposed approach.
{"title":"Improving Symbolic Regression through a semantics-driven framework","authors":"Q. Huynh, H. Singh, T. Ray","doi":"10.1109/SSCI.2016.7849941","DOIUrl":"https://doi.org/10.1109/SSCI.2016.7849941","url":null,"abstract":"The process of identifying analytical relationships among variables and responses in observed data is commonly referred to as Symbolic Regression (SR). Genetic Programming is one of the commonly used approaches for SR, which operates by evolving expressions. Such relationships could be explicit or implicit in nature, of which the former has been more extensively studied in literature. Even though extensive studies have been done in SR, the fundamental challenges such as bloat, loss of diversity and accurate determination of coefficients still persist. Recently, semantics and multi-objective formulation have been suggested as potential tools to alleviate these issues by building more intelligence in the search process. However, studies along both these directions have been in isolation and applied only to selected components of SR so far. In this paper, we intend to build a framework that integrates semantics deeper into more components of SR. The framework could be operated in conventional single objective as well as multi-objective mode and is capable of dealing with both explicit and implicit functions. Semantics are used in the proposed framework for improving compactness and diversity of expressions, crossover and local exploitation. Numerical experiments are presented on a set of benchmark problems to demonstrate the strengths of the proposed approach.","PeriodicalId":120288,"journal":{"name":"2016 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2016-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116171966","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 : 2016-12-01DOI: 10.1109/SSCI.2016.7850217
Maroi Agrebi, M. Abed, Mohamed Nazih Omri
The location selection of distribution centers is one of important strategies to optimize the logistics system. To solve this problem, this paper presents a new multi-actor multi-attribute decision-making method based on ELECTRE I. The proposed method helps decision-makers to select a preferred location from a given set of locations for implementing. The strength of the proposed method is to incorporate the preferences of a set of decision-makers into account, notably the role of their experience into the decision-making process, consider both quantitative and qualitative criteria, take into account both desirable directions (Min and Max) and validate the selected location by both tests of concordance and discordance simultaneously. A case study is provided to illustrate the proposed method.
{"title":"A new multi-actor multi-attribute decision-making method to select the distribution centers' location","authors":"Maroi Agrebi, M. Abed, Mohamed Nazih Omri","doi":"10.1109/SSCI.2016.7850217","DOIUrl":"https://doi.org/10.1109/SSCI.2016.7850217","url":null,"abstract":"The location selection of distribution centers is one of important strategies to optimize the logistics system. To solve this problem, this paper presents a new multi-actor multi-attribute decision-making method based on ELECTRE I. The proposed method helps decision-makers to select a preferred location from a given set of locations for implementing. The strength of the proposed method is to incorporate the preferences of a set of decision-makers into account, notably the role of their experience into the decision-making process, consider both quantitative and qualitative criteria, take into account both desirable directions (Min and Max) and validate the selected location by both tests of concordance and discordance simultaneously. A case study is provided to illustrate the proposed method.","PeriodicalId":120288,"journal":{"name":"2016 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2016-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122781407","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 : 2016-12-01DOI: 10.1109/SSCI.2016.7850212
Xuhua Shi, Chudong Tong, Li Wang
Surrogate modelling and model management are key points for evolutionary optimization of chemical processes. This paper proposes an evolutionary algorithm with the help of adaptive surrogate functions (EASF), in which approximate models' establishment and management are combined to search the optimal result. To construct an appropriate surrogate model, a new hybrid modelling framework with adaptive Radial Basis Functions (RBF) (ARBF) is put forward. Different from most neural network modelling methods, ARBF is able to adaptively adjust the sample size by current approximation errors to effectively take into account the tradeoff between approximation accuracy and sample size. For model management, an approximation error fuzzy control strategy (AEFCS) is introduced. AEFCS in combination with ARBF can effectively perform exploratory and exploitative search in the evolutionary optimization. The superiority of EASF is demonstrated by the simulation results on three benchmark problems. To illustrate the performance of EASF further, it is employed to optimize the operating conditions of crude oil distillation process, and satisfactory results are obtained.
{"title":"Evolutionary optimization with adaptive surrogates and its application in crude oil distillation","authors":"Xuhua Shi, Chudong Tong, Li Wang","doi":"10.1109/SSCI.2016.7850212","DOIUrl":"https://doi.org/10.1109/SSCI.2016.7850212","url":null,"abstract":"Surrogate modelling and model management are key points for evolutionary optimization of chemical processes. This paper proposes an evolutionary algorithm with the help of adaptive surrogate functions (EASF), in which approximate models' establishment and management are combined to search the optimal result. To construct an appropriate surrogate model, a new hybrid modelling framework with adaptive Radial Basis Functions (RBF) (ARBF) is put forward. Different from most neural network modelling methods, ARBF is able to adaptively adjust the sample size by current approximation errors to effectively take into account the tradeoff between approximation accuracy and sample size. For model management, an approximation error fuzzy control strategy (AEFCS) is introduced. AEFCS in combination with ARBF can effectively perform exploratory and exploitative search in the evolutionary optimization. The superiority of EASF is demonstrated by the simulation results on three benchmark problems. To illustrate the performance of EASF further, it is employed to optimize the operating conditions of crude oil distillation process, and satisfactory results are obtained.","PeriodicalId":120288,"journal":{"name":"2016 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2016-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114462057","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 : 2016-12-01DOI: 10.1109/SSCI.2016.7849925
Alexander Dockhorn, Christian Braune, R. Kruse
The class of density-based clustering algorithms excels in detecting clusters of arbitrary shape. DBSCAN, the most common representative, has been demonstrated to be useful in a lot of applications. Still the algorithm suffers from two drawbacks, namely a non-trivial parameter estimation for a given dataset and the limitation to data sets with constant cluster density. The first was already addressed in our previous work, where we presented two hierarchical implementations of DBSCAN. In combination with a simple optimization procedure, those proofed to be useful in detecting appropriate parameter estimates based on an objective function. However, our algorithm was not capable of producing clusters of differing density. In this work we will use the hierarchical information to extract variable density clusters and nested cluster structures. Our evaluation shows that the clustering approach based on edge-lengths of the dendrogram or based on area estimates successfully detects clusters of arbitrary shape and density.
{"title":"Variable density based clustering","authors":"Alexander Dockhorn, Christian Braune, R. Kruse","doi":"10.1109/SSCI.2016.7849925","DOIUrl":"https://doi.org/10.1109/SSCI.2016.7849925","url":null,"abstract":"The class of density-based clustering algorithms excels in detecting clusters of arbitrary shape. DBSCAN, the most common representative, has been demonstrated to be useful in a lot of applications. Still the algorithm suffers from two drawbacks, namely a non-trivial parameter estimation for a given dataset and the limitation to data sets with constant cluster density. The first was already addressed in our previous work, where we presented two hierarchical implementations of DBSCAN. In combination with a simple optimization procedure, those proofed to be useful in detecting appropriate parameter estimates based on an objective function. However, our algorithm was not capable of producing clusters of differing density. In this work we will use the hierarchical information to extract variable density clusters and nested cluster structures. Our evaluation shows that the clustering approach based on edge-lengths of the dendrogram or based on area estimates successfully detects clusters of arbitrary shape and density.","PeriodicalId":120288,"journal":{"name":"2016 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2016-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114495982","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 : 2016-12-01DOI: 10.1109/SSCI.2016.7850157
Yiqi Deng, P. Bentley, Alvee Momshad
Biological systems have become highly significant for traditional computer architectures as examples of highly complex self-organizing systems that perform tasks in parallel with no centralized control. However, few researchers have compared the suitability of different computing approaches for the unique features of Artificial Immune Systems (AIS) when trying to introduce novel computing architectures, and few consider the practicality of their solutions for real world machine learning problems. We propose that the efficacy of AIS-based computing for tackling real world datasets can be improved by the exploitation of intrinsic features of computer architectures. This paper reviews and evaluates current existing implementation solutions for AIS on different computing paradigms and introduces the idea of “C Principles” and “A Principles”. Three Artificial Immune Systems implemented on different architectures are compared using these principles to examine the possibility of improving AIS through taking advantage of intrinsic hardware features.
{"title":"Improving Artificial-Immune-System-based computing by exploiting intrinsic features of computer architectures","authors":"Yiqi Deng, P. Bentley, Alvee Momshad","doi":"10.1109/SSCI.2016.7850157","DOIUrl":"https://doi.org/10.1109/SSCI.2016.7850157","url":null,"abstract":"Biological systems have become highly significant for traditional computer architectures as examples of highly complex self-organizing systems that perform tasks in parallel with no centralized control. However, few researchers have compared the suitability of different computing approaches for the unique features of Artificial Immune Systems (AIS) when trying to introduce novel computing architectures, and few consider the practicality of their solutions for real world machine learning problems. We propose that the efficacy of AIS-based computing for tackling real world datasets can be improved by the exploitation of intrinsic features of computer architectures. This paper reviews and evaluates current existing implementation solutions for AIS on different computing paradigms and introduces the idea of “C Principles” and “A Principles”. Three Artificial Immune Systems implemented on different architectures are compared using these principles to examine the possibility of improving AIS through taking advantage of intrinsic hardware features.","PeriodicalId":120288,"journal":{"name":"2016 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2016-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121914791","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 : 2016-12-01DOI: 10.1109/SSCI.2016.7850200
O. Shadura, F. Carminati
Machine learning for complex multi-objective problems (MOP) can substantially speedup the discovery of solutions belonging to Pareto landscapes and improve Pareto front accuracy. Studying convergence speedup of multi-objective search on well-known benchmarks is an important step in the development of algorithms to optimize complex problems such as High Energy Physics particle transport simulations. In this paper we will describe how we perform this optimization via a tuning based on genetic algorithms and machine learning for MOP. One of the approaches described is based on the introduction of a specific multivariate analysis operator that can be used in case of expensive fitness function evaluations, in order to speed-up the convergence of the “black-box” optimization problem.
{"title":"Stochastic performance tuning of complex simulation applications using unsupervised machine learning","authors":"O. Shadura, F. Carminati","doi":"10.1109/SSCI.2016.7850200","DOIUrl":"https://doi.org/10.1109/SSCI.2016.7850200","url":null,"abstract":"Machine learning for complex multi-objective problems (MOP) can substantially speedup the discovery of solutions belonging to Pareto landscapes and improve Pareto front accuracy. Studying convergence speedup of multi-objective search on well-known benchmarks is an important step in the development of algorithms to optimize complex problems such as High Energy Physics particle transport simulations. In this paper we will describe how we perform this optimization via a tuning based on genetic algorithms and machine learning for MOP. One of the approaches described is based on the introduction of a specific multivariate analysis operator that can be used in case of expensive fitness function evaluations, in order to speed-up the convergence of the “black-box” optimization problem.","PeriodicalId":120288,"journal":{"name":"2016 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2016-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122098207","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 : 2016-12-01DOI: 10.1109/SSCI.2016.7850233
C. Wong, S. Sundaram
DIRECT is known for balancing the exploration and exploitation of a search space. This paper seeks to explore the improvement of diversity among solutions through the use of qualitative indicators in multi-objective DIRECT framework. Three different indicators - Hypervolume (HV), Epsilon (EPS), R2 indicators are used in this study. The three variants of indicators are tested on the Black-box Multi-objective Optimization Benchmarking (BMOB) Platform. The results are presented and some insights in the choice of selection operator are provided. Overall, HV indicator performs the best followed by R2, then EPS. EPS indicator performs worse than HV and R2 in unimodal problems. Also, HV indicator achieves notably better results at high dimensions. R2 performs better than EPS in non-separable problems.
{"title":"Preliminary study: Qualitative indicators in multi-objective DIRECT framework","authors":"C. Wong, S. Sundaram","doi":"10.1109/SSCI.2016.7850233","DOIUrl":"https://doi.org/10.1109/SSCI.2016.7850233","url":null,"abstract":"DIRECT is known for balancing the exploration and exploitation of a search space. This paper seeks to explore the improvement of diversity among solutions through the use of qualitative indicators in multi-objective DIRECT framework. Three different indicators - Hypervolume (HV), Epsilon (EPS), R2 indicators are used in this study. The three variants of indicators are tested on the Black-box Multi-objective Optimization Benchmarking (BMOB) Platform. The results are presented and some insights in the choice of selection operator are provided. Overall, HV indicator performs the best followed by R2, then EPS. EPS indicator performs worse than HV and R2 in unimodal problems. Also, HV indicator achieves notably better results at high dimensions. R2 performs better than EPS in non-separable problems.","PeriodicalId":120288,"journal":{"name":"2016 IEEE Symposium Series on Computational Intelligence (SSCI)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2016-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116833257","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：