scispace - formally typeset
Search or ask a question
Author

M. Costa

Bio: M. Costa is an academic researcher from Polytechnic University of Turin. The author has contributed to research in topics: Sorting & Estimation of distribution algorithm. The author has an hindex of 4, co-authored 8 publications receiving 145 citations.

Papers
More filters
Journal Article
TL;DR: In this article, an evolutionary multi-objective optimization tool based on an estimation of distribution algorithm is proposed, which uses the ranking method of non-dominated sorting genetic algorithm-II and the Parzen estimator to approximate the probability density of solutions lying on the Pareto front.
Abstract: An evolutionary multi-objective optimization tool based on an estimation of distribution algorithm is proposed. The algorithm uses the ranking method of non-dominated sorting genetic algorithm-II and the Parzen estimator to approximate the probability density of solutions lying on the Pareto front. The proposed algorithm has been applied to different types of test case problems and results show good performance of the overall optimization procedure in terms of the number of function evaluations. An alternative spreading technique that uses the Parzen estimator in the objective function space is proposed as well. When this technique is used, achieved results appear to be qualitatively equivalent to those previously obtained by adopting the crowding distance described in non-dominated sorting genetic algorithm-II.

66 citations

Book ChapterDOI
08 Apr 2003
TL;DR: An evolutionary multi-objective optimization tool based on an estimation of distribution algorithm and the Parzen estimator that appears to be qualitatively equivalent to those previously obtained by adopting the crowding distance described in non-dominated sorting genetic algorithm-II.
Abstract: An evolutionary multi-objective optimization tool based on an estimation of distribution algorithm is proposed. The algorithm uses the ranking method of non-dominated sorting genetic algorithm-II and the Parzen estimator to approximate the probability density of solutions lying on the Pareto front. The proposed algorithm has been applied to different types of test case problems and results show good performance of the overall optimization procedure in terms of the number of function evaluations. An alternative spreading technique that uses the Parzen estimator in the objective function space is proposed as well. When this technique is used, achieved results appear to be qualitatively equivalent to those previously obtained by adopting the crowding distance described in non-dominated sorting genetic algorithm-II.

60 citations

Proceedings ArticleDOI
19 May 2001
TL;DR: The authors extend the ISO/TC213 partitioning to probability density functions so as to include the measurement process in the formalism and make use of unsupervised probabilistic neural networks to build a semi-parametric Probabilistic model for each class of symmetry.
Abstract: The traditional approach to the geometrical dimensioning and tolerancing of mechanical components and assemblies essentially relies on definitions by examples. Over the last few years that approach is increasingly being challenged by a unifying and theoretically sound perspective. The technical commission ISO/TC213 devised a very elegant and powerful classification of 3D objects based on their symmetries. The authors embed that classification in a fully fledged probabilistic framework and propose a practical methodology for the statistical recognition of 3D shapes from sparse, noisy measurements. To this purpose we first extend the ISO/TC213 partitioning to probability density functions so as to include the measurement process in the formalism. Then we make use of unsupervised probabilistic neural networks to build a semi-parametric probabilistic model for each class of symmetry. Finally, we rank all competing models against clouds of measured points according to their leave-one-out likelihood.

10 citations

Book ChapterDOI
04 Jun 2003
TL;DR: An evolutionary multi-objective optimization tool based on an estimation of distribution algorithm that uses the ranking method of non-dominated sorting genetic algorithm-II and the Parzen estimator to approximate the probability density of solutions lying on the Pareto front is proposed.
Abstract: Evolutionary algorithms perform optimization using the information derived from a population of sample solution points. Recent developments in this field regard optimization as the evolutionary process of an explicit, probabilistic model of the search space. The algorithms derived on the basis of this new philosophy maintain every feature of the classic evolutionary algorithms, but are able to overcome some drawbacks. In this paper an evolutionary multi-objective optimization tool based on an estimation of distribution algorithm is proposed. It uses the ranking method of non-dominated sorting genetic algorithm-II and the Parzen estimator to approximate the probability density of solutions lying on the Pareto front. The proposed algorithm has been applied to different types of test case problems and results show good performance of the overall optimization procedure in terms of the number of function evaluations.

7 citations

01 Jan 2002
TL;DR: Present technology provides the tools to implement the probabilistic approach, as demonstrated by some examples, and in the next future it could be the new standard in the definition of product shape.
Abstract: The definition, management and control of geometrical shape affect the functionality of industrial product. A unified approach to the description of the product shape along the design, manufacturing and inspection phases can be provided by a probabilistic interpretation of product boundaries. Present technology provides the tools to implement the probabilistic approach, as demonstrated by some examples, and in the next future it could be the new standard in the definition of product shape.

3 citations


Cited by
More filters
Proceedings ArticleDOI
Aimin Zhou1, Yaochu Jin2, Qingfu Zhang1, Bernhard Sendhoff2, Edward Tsang1 
24 Jan 2006
TL;DR: The proposed hybrid method is verified on widely used test problems and simulation results show that the method is effective in achieving Pareto-optimal solutions compared to two state-of-the-art evolutionary multi-objective algorithms.
Abstract: In our previous work conducted by Aimin Zhou et. al., (2005), it has been shown that the performance of multi-objective evolutionary algorithms can be greatly enhanced if the regularity in the distribution of Pareto-optimal solutions is used. This paper suggests a new hybrid multi-objective evolutionary algorithm by introducing a convergence based criterion to determine when the model-based method and when the genetics-based method should be used to generate offspring in each generation. The basic idea is that the genetics-based method, i.e., crossover and mutation, should be used when the population is far away from the Pareto front and no obvious regularity in population distribution can be observed. When the population moves towards the Pareto front, the distribution of the individuals will show increasing regularity and in this case, the model-based method should be used to generate offspring. The proposed hybrid method is verified on widely used test problems and our simulation results show that the method is effective in achieving Pareto-optimal solutions compared to two state-of-the-art evolutionary multi-objective algorithms: NSGA-II and SPEA2, and our pervious method in Aimin Zhou et. al., (2005).

281 citations

Journal ArticleDOI
TL;DR: This paper proposes a new model-based method for representing and searching nondominated solutions that is able to alleviate the requirement on solution diversity and in principle, as many solutions as needed can be generated.
Abstract: To approximate the Pareto front, most existing multiobjective evolutionary algorithms store the nondominated solutions found so far in the population or in an external archive during the search. Such algorithms often require a high degree of diversity of the stored solutions and only a limited number of solutions can be achieved. By contrast, model-based algorithms can alleviate the requirement on solution diversity and in principle, as many solutions as needed can be generated. This paper proposes a new model-based method for representing and searching nondominated solutions. The main idea is to construct Gaussian process-based inverse models that map all found nondominated solutions from the objective space to the decision space. These inverse models are then used to create offspring by sampling the objective space. To facilitate inverse modeling, the multivariate inverse function is decomposed into a group of univariate functions, where the number of inverse models is reduced using a random grouping technique. Extensive empirical simulations demonstrate that the proposed algorithm exhibits robust search performance on a variety of medium to high dimensional multiobjective optimization test problems. Additional nondominated solutions are generated a posteriori using the constructed models to increase the density of solutions in the preferred regions at a low computational cost.

248 citations

BookDOI
01 Jan 2012
TL;DR: This work proposes customized model ensembles on demand, inspired by Lazy Learning, which finds the most relevant models from a DB of models, using their meta-information, and creates an ensemble, which produces an output that is a weighted interpolation or extrapolation of the outputs of the models ensemble.
Abstract: In the not so distant future, we expect analytic models to become a commodity. We envision having access to a large number of data-driven models, obtained by a combination of crowdsourcing, crowdservicing, cloud-based evolutionary algorithms, outsourcing, in-house development, and legacy models. In this new context, the critical question will be model ensemble selection and fusion, rather than model generation. We address this issue by proposing customized model ensembles on demand, inspired by Lazy Learning. In our approach, referred to as Lazy Meta-Learning, for a given query we find the most relevant models from a DB of models, using their meta-information. After retrieving the relevant models, we select a subset of models with highly uncorrelated errors. With these models we create an ensemble and use their meta-information for dynamic bias compensation and relevance weighting. The output is a weighted interpolation or extrapolation of the outputs of the models ensemble. Furthermore, the confidence interval around the output is reduced as we increase the number of uncorrelated models in the ensemble. We have successfully tested this approach in a power plant management application.

171 citations

Journal ArticleDOI
TL;DR: A taxonomy and a comprehensive review of applications of MOEAs in aeronautical and aerospace design problems are presented and some potential paths for future research are provided, which are considered promising within this area.
Abstract: Nowadays, the solution of multiobjective optimization problems in aeronautical and aerospace engineering has become a standard practice. These two fields offer highly complex search spaces with different sources of difficulty, which are amenable to the use of alternative search techniques such as metaheuristics, since they require little domain information to operate. From the several metaheuristics available, multiobjective evolutionary algorithms (MOEAs) have become particularly popular, mainly because of their availability, ease of use, and flexibility. This paper presents a taxonomy and a comprehensive review of applications of MOEAs in aeronautical and aerospace design problems. The review includes both the characteristics of the specific MOEA adopted in each case, as well as the features of the problems being solved with them. The advantages and disadvantages of each type of approach are also briefly addressed. We also provide a set of general guidelines for using and designing MOEAs for aeronautical and aerospace engineering problems. In the final part of the paper, we provide some potential paths for future research, which we consider promising within this area.

142 citations

Journal ArticleDOI
TL;DR: This paper shows how probabilistic graphical models have been used in evolutionary algorithms to improve their performance in solving complex problems and gives a survey of Probabilistic model building-based evolutionary algorithms, called estimation of distribution algorithms, and compares different methods for probabilism modeling in these algorithms.
Abstract: Thanks to their inherent properties, probabilistic graphical models are one of the prime candidates for machine learning and decision making tasks especially in uncertain domains. Their capabilities, like representation, inference and learning, if used effectively, can greatly help to build intelligent systems that are able to act accordingly in different problem domains. Evolutionary algorithms is one such discipline that has employed probabilistic graphical models to improve the search for optimal solutions in complex problems. This paper shows how probabilistic graphical models have been used in evolutionary algorithms to improve their performance in solving complex problems. Specifically, we give a survey of probabilistic model building-based evolutionary algorithms, called estimation of distribution algorithms, and compare different methods for probabilistic modeling in these algorithms.

100 citations