Results 1  10
of
12
Parallel estimation of distribution algorithms
, 2002
"... The thesis deals with the new evolutionary paradigm based on the concept of Estimation of Distribution Algorithms (EDAs) that use probabilistic model of promising solutions found so far to obtain new candidate solutions of optimized problem. There are six primary goals of this thesis: 1. Suggestion ..."
Abstract

Cited by 26 (4 self)
 Add to MetaCart
The thesis deals with the new evolutionary paradigm based on the concept of Estimation of Distribution Algorithms (EDAs) that use probabilistic model of promising solutions found so far to obtain new candidate solutions of optimized problem. There are six primary goals of this thesis: 1. Suggestion of a new formal description of EDA algorithm. This high level concept can be used to compare the generality of various probabilistic models by comparing the properties of underlying mappings. Also, some convergence issues are discussed and theoretical ways for further improvements are proposed. 2. Development of new probabilistic model and methods capable of dealing with continuous parameters. The resulting Mixed Bayesian Optimization Algorithm (MBOA) uses a set of decision trees to express the probability model. Its main advantage against the mostly used IDEA and EGNA approach is its backward compatibility with discrete domains, so it is uniquely capable of learning linkage between mixed continuousdiscrete genes. MBOA handles the discretization of continuous parameters as an integral part of the learning process, which outperforms the histogrambased
Matching inductive search bias and problem structure in continuous estimation of distribution algorithms
 European Journal of Operational Research
"... Research into the dynamics of Genetic Algorithms (GAs) has led to the ¯eld of Estimation{of{Distribution Algorithms (EDAs). For discrete search spaces, EDAs have been developed that have obtained very promising results on a wide variety of problems. In this paper we investigate the conditions under ..."
Abstract

Cited by 16 (3 self)
 Add to MetaCart
(Show Context)
Research into the dynamics of Genetic Algorithms (GAs) has led to the ¯eld of Estimation{of{Distribution Algorithms (EDAs). For discrete search spaces, EDAs have been developed that have obtained very promising results on a wide variety of problems. In this paper we investigate the conditions under which the adaptation of this technique to continuous search spaces fails to perform optimization e±ciently. We show that without careful interpretation and adaptation of lessons learned from discrete EDAs, continuous EDAs will fail to perform e±cient optimization on even some of the simplest problems. We reconsider the most important lessons to be learned in the design of EDAs and subsequently show how we can use this knowledge to extend continuous EDAs that were obtained by straightforward adaptation from the discrete domain so as to obtain an improvement in performance. Experimental results are presented to illustrate this improvement and to additionally con¯rm experimentally that a proper adaptation of discrete EDAs to the continuous case indeed requires careful consideration. Key words: Estimation{of{distribution algorithms; Numerical optimization;
Experimental Study: Hypergraph Partitioning Based on the Simple and Advanced Genetic Algorithm BMDA and
 In Proceedings of the Fifth International Conference on Soft Computing
, 1999
"... Abstract: This paper is an experimental study on hypegraph partitioning using the simple genetic algorithm (GA) based on the schema theorem and the advanced algorithms based on the estimation of distribution of promising solution. Primarily we have implemented a simple GA based on the GaLib library[ ..."
Abstract

Cited by 11 (4 self)
 Add to MetaCart
(Show Context)
Abstract: This paper is an experimental study on hypegraph partitioning using the simple genetic algorithm (GA) based on the schema theorem and the advanced algorithms based on the estimation of distribution of promising solution. Primarily we have implemented a simple GA based on the GaLib library[Gal94] and some hybrid variant included a fast heuristics to speed up the convergence of the optimization process. Secondly we have implemented the Univariate Marginal Distribution algorithm (UMDA) and the Bivariate Marginal Distribution algorithm (BMDA), both have been published even recently[Pel98] and used a share version of a superior new program BOA based on the Bayesian Optimization Algorithm [Pel99]. We have also extended the BMDA algorithm to a new version with finite alphabet encoding of chromozomes and new metric that enables the mway partitioning graphs. The aim of our paper is to test the efficiency of new approaches for discrete combinatorial problems represented by hypergraph partitioning. Key words: decomposition, hypergraph partitioning, simple and hybrid GA, estimation of distribution algorithm, Bayesian network. 1
Automatic discovery of ranking formulas for playing with multiarmed
"... bandits ..."
(Show Context)
iECGA: Integer extended compact genetic algorithm
 IN PROCEEDINGS OF ACM SIGEVO GENETIC AND EVOLUTIONARY COMPUTATION CONFERENCE (GECCO2006
, 2006
"... Extended compact genetic algorithm (ECGA) is an algorithm that can solve hard problems in the binary domain. ECGA is reliable and accurate because of the capability of detecting building blocks, but certain difficulties are encountered when we directly apply ECGA to problems in the integer domain. I ..."
Abstract

Cited by 5 (2 self)
 Add to MetaCart
(Show Context)
Extended compact genetic algorithm (ECGA) is an algorithm that can solve hard problems in the binary domain. ECGA is reliable and accurate because of the capability of detecting building blocks, but certain difficulties are encountered when we directly apply ECGA to problems in the integer domain. In this paper, we propose a new algorithm that extends ECGA, called integer extended compact genetic algorithm (iECGA). iECGA uses a modified probability model and inherits the capability of detecting building blocks from ECGA. iECGA is specifically designed for problems in the integer domain and can avoid the difficulties that ECGA encounters. In the experimental results, we show the performance comparisons between ECGA, iECGA, and a simple GA, and the results indicate that iECGA has good performances on problems in the integer domain.
Multiobjective Bayesian Optimization Algorithm for Combinatorial Problems: Theory and practice
"... This paper deals with the utilizing of the Bayesian optimization algorithm (BOA) for the multiobjective optimization of combinatorial problems. Three probabilistic models used in the Estimation Distribution Algorithms (EDA), such as UMDA, BMDA and BOA which allow to search effectively on the promi ..."
Abstract

Cited by 3 (1 self)
 Add to MetaCart
This paper deals with the utilizing of the Bayesian optimization algorithm (BOA) for the multiobjective optimization of combinatorial problems. Three probabilistic models used in the Estimation Distribution Algorithms (EDA), such as UMDA, BMDA and BOA which allow to search effectively on the promising areas of the combinatorial search space are discussed. The main attention is focused on the incorporation of Pareto optimality concept into classical structure of the BOA algorithm. We have modified the standard algorithm BOA for one criterion optimization utilizing the known niching techniques to find the Pareto optimal set. The experiments are focused on tree classes of the combinatorial problems: artificial problem with known Pareto set, multiple 0/1 knapsack problem and the bisectioning of hypergraphs as well.
Optimized lookahead tree search policies
"... Abstract. We consider in this paper lookahead tree techniques for the discretetime control of a deterministic dynamical system so as to maximize a sum of discounted rewards over an infinite time horizon. Given the current system state xt at time t, these techniques explore the lookahead tree repre ..."
Abstract

Cited by 3 (3 self)
 Add to MetaCart
(Show Context)
Abstract. We consider in this paper lookahead tree techniques for the discretetime control of a deterministic dynamical system so as to maximize a sum of discounted rewards over an infinite time horizon. Given the current system state xt at time t, these techniques explore the lookahead tree representing possible evolutions of the system states and rewards conditioned on subsequent actions ut, ut+1,.... When the computing budget is exhausted, they output the action ut that led to the best found sequence of discounted rewards. In this context, we are interested in computing good strategies for exploring the lookahead tree. We propose a generic approach that looks for such strategies by solving an optimization problem whose objective is to compute a (budget compliant) treeexploration strategy yielding a control policy maximizing the average return over a postulated set of initial states. This generic approach is fully specified to the case where the space of candidate treeexploration strategies are “bestfirst ” strategies parameterized by a linear combination of lookahead path features – some of them having been advocated in the literature before – and where the optimization problem is solved by using an EDAalgorithm based on Gaussian distributions. Numerical experiments carried out on a model of the treatment of the HIV infection show that the optimized treeexploration strategy is orders of magnitudes better than the previously advocated ones.
Evoptool: an Extensible Toolkit for Evolutionary Optimization Algorithms Comparison
"... Abstract — This paper presents Evolutionary Optimization Tool (Evoptool), an optimization toolkit that implements a set of metaheuristics based on the Evolutionary Computation paradigm. Evoptool provides a common platform for the development and test of new algorithms, in order to facilitate the pe ..."
Abstract

Cited by 3 (3 self)
 Add to MetaCart
(Show Context)
Abstract — This paper presents Evolutionary Optimization Tool (Evoptool), an optimization toolkit that implements a set of metaheuristics based on the Evolutionary Computation paradigm. Evoptool provides a common platform for the development and test of new algorithms, in order to facilitate the performance comparison activity. The toolkit offers a wide set of benchmark problems, from classical toy examples to complex tasks, and a collection of implementations of algorithms from the Genetic Algorithms and Estimation of Distribution Algorithms paradigms. Evoptool is flexible and easy to extend, also with algorithms based on other approaches that go beyond Evolutionary Computation. I.
Metalearning of exploration/exploitation strategies: The multiarmed bandit case
 in Proc. Int. Conf. Agents Artif. Intell., 2012 [Online]. Available: arXiv:1207.5208
"... Abstract. The exploration/exploitation (E/E) dilemma arises naturally in many subfields of Science. Multiarmed bandit problems formalize this dilemma in its canonical form. Most current research in this field focuses on generic solutions that can be applied to a wide range of problems. However, in ..."
Abstract

Cited by 1 (1 self)
 Add to MetaCart
(Show Context)
Abstract. The exploration/exploitation (E/E) dilemma arises naturally in many subfields of Science. Multiarmed bandit problems formalize this dilemma in its canonical form. Most current research in this field focuses on generic solutions that can be applied to a wide range of problems. However, in practice, it is often the case that a form of prior information is available about the specific class of target problems. Prior knowledge is rarely used in current solutions due to the lack of a systematic approach to incorporate it into the E/E strategy. To address a specific class of E/E problems, we propose to proceed in three steps: (i) model prior knowledge in the form of a probability distribution over the target class of E/E problems; (ii) choose a large hypothesis space of candidate E/E strategies; and (iii), solve an optimization problem to find a candidate E/E strategy of maximal average performance over a sample of problems drawn from the prior distribution. We illustrate this metalearning approach with two different hypothesis spaces: one where E/E strategies are numerically parameterized and another where E/E strategies are represented as small symbolic formulas. We propose appropriate optimization algorithms for both cases. Our experiments, with twoarmed “Bernoulli ” bandit problems and various playing budgets, show that the metalearnt E/E strategies outperform generic strategies of the literature (UCB1, UCB1TUNED, UCBV, KLUCB and nGREEDY); they also evaluate the robustness of the learnt E/E strategies, by tests carried out on arms whose rewards follow a truncated Gaussian distribution.
A Novel Approach to Model Selection in Distribution Estimation Using Markov Networks
, 2011
"... Evolutionary Algorithms consist of a broad class of optimization algorithms based on the Darwinian paradigm of the survival of the fittest. In particular, Estimation of Distribution Algorithms (EDAs) are a recent paradigm, often presented in the literature as an evolution of Genetic Algorithms, th ..."
Abstract
 Add to MetaCart
Evolutionary Algorithms consist of a broad class of optimization algorithms based on the Darwinian paradigm of the survival of the fittest. In particular, Estimation of Distribution Algorithms (EDAs) are a recent paradigm, often presented in the literature as an evolution of Genetic Algorithms, that integrates the evolutionary approach with techniques from Statistical Machine Learning. An EDA evolves a population of promising solutions to a given optimization problem by building and sampling probabilistic models. This thesis proposes to integrate EDAs based on Markov Networks with novel model selection techniques from Statistical Machine Learning community. Particular focus is given to algorithms in the DEUM framework, a recent promising searchstrategy based on undirected graphical models. We introduce the use of `1regularized logistic regression techniques with the aim to recover the interactions among the variables of a given optimization problem