1,721,020 research outputs found
Differential evolution and particle swarm optimisation in partitional clustering
No full textMany partitional clustering algorithms based on genetic algorithms (GA) have been proposed to tackle the problem of finding the optimal partition of a data set. Very few studies considered alternative stochastic search heuristics other than GAs or simulated annealing. Two promising algorithms for numerical optimisation, which are hardly known outside the search heuristics field, are particle swarm optimisation (PSO) and differential evolution (DE). The performance of GAs for a representative point evolution approach to clustering is compared with PSO and DE. The empirical results show that DE is clearly and consistently superior compared to GAs and PSO for hard clustering problems, both with respect to precision as well as robustness (reproducibility) of the results. Only for simple data sets, the GA and PSO can obtain the same quality of results. Apart from superior performance, DE is easy to implement and requires hardly any parameter tuning compared to substantial tuning for GAs and PSOs. Our study shows that DE rather than GAs should receive primary attention in partitional clustering algorithms
Multiobjective Optimization using Differential Evolution for Real-World Portfolio Optimization
No full textPortfolio optimization is an important aspect of decision-support in investment management. Realistic portfolio optimization, in contrast to simplistic mean- variance optimization, is a challenging problem, because it requires to determine a setof optimal solutions with respect to multiple objectives, where the objective functions are often multimodal and non-smooth. Moreover, the objectives are subject to various constraints of which many are typically non-linear and discontinuous. Conventional optimization methods, such as quadratic programming, cannot cope with these realistic problem properties. A valuable alternative are stochastic search heuristics, such as simulated annealing or evolutionary algorithms. We propose a new multiobjective evolutionary algorithm for portfolio optimization, which we call DEMPO - Differential Evolution for Multiobjective Portfolio Optimization. In our experimentation, we compare DEMPO with quadratic programming and another well-known evolutionary algorithm for multiobjective optimization called NSGA-II. The main advantage of DEMPO is its ability to tackle a portfolio optimization task without simplications, while obtaining very satisfying results in reasonable runtime
Additive modeling for location, scale, and shape parameters of the skew normal distribution
No full textIn questo lavoro, si propone di modellare additivamente i parametri di locazione, scala e simmetria della normale asimmetrica (Azzalini (2005)), estendendo l’approcciodi Rigby and Stasinopoulos (2005) a tale classe di distribuzioni. Il comportamento del modello ́e stato dapprima testato empiricamente mediante simulazioni Monte Carlo, verificando la robustezza dell’algoritmo di ottimizzazione utilizzato (differentialevolution) e la stabilit ́a nella stima dei parametri. Infine, tale modello `e stato utilizzato per analizzare dati relativi all’evoluzione temporale della capacit`a cranica umana
The Maximum Lq-Likelihood Estimator in Extreme Value Theory, Italian
No full textIn questo lavoro, si propone di utilizzare lo stimatore MLqE di Massima Lq-Verosimiglianza, introdotto da Ferrari e Yang (2007), per la stima dei parametri della distribuzione del Valore Estremo generalizzata e della distribuzione di Pareto generaliz- zata. L’analisi empirica, condotta mediante simulazioni Monte Carlo, mostra che lo sti- matore MLqE e ́ piu ́ efficiente dello stimatore di Massima Verosimiglianza nel caso in cui si voglia stimare la probabilita ́ di un evento estremo, avendo a disposizione un campione di dimensioni limitate
Evolutionary Clustering Analysis
No full textThe determination of the number of groups in a dataset, theircomposition and the most relevant measurements to be considered in clusteringthe data, is a high-demanding task, especially when the a priori information onthe dataset is limited. Three different genetic approaches are introduced in thispaper as tools for automatic data clustering and features selection. They differin the adopted codification of the grouping problem, not in the evolutionaryoperator and parameters. Two of them deals with the grouping problem in adeterministic framework. The first directly approaches the grouping problem asa combinatorial one. The second tries to determine some relevant points in thedata domain to be used in clustering data as group separators. A probabilisticframework is then introduced with the third one, which starts specifying thestatistical model from which data are assumed to be drawn. The evolutionaryapproaches are, finally, compared with respect to classical partitional clusteringalgorithms on simulated data and on Fisher’s Iris dataset used as a benchmark
High Performance Clustering with Differential Evolution
No full textPartitional clustering poses a NP hard search problem for non-trivial problems. While genetic algorithms (GA) have been very popular in the clustering field, particle swarm optimization (PSO) and differential evolution (DE) are rather unknown. In this paper, we report results of a performance comparison between a GA, PSO and DE for a medoid evolution clusterign approach. Our results show that DE is clearly and consistently superior compared to FAs and PSO. both in respect to precision and robustness of the results for hard clustering problems. We conclude that DE rather than GAs should be primarily considered for tackling partitional clustering problems with numerical optimization
Evolutionary Approaches for Statistical Modelling
No full textIn this paper, we describe some evolutionaryapproaches based on genetic algorithms to face the statisticalmodel selection problem using completely data-drivenalgorithms. As first, we propose an approach to selectmultivariate linear regression models as well as to buildARMA time series models. As second, we introduce amethodology to tackle the clustering problem in a modelbasedframework. We report the results from severalapplications and from simulated datasets and we compare theevolutionary approaches with some classical ones
Book Review
No full textBook ReviewManfred Gilli, Dietmar Maringer & Enrico Schumann, “Numerical Methods and Optimization in Finance”, Academic Press, 2011, 600 pages, USD 99.95, ISBN: 978-0-12-375662-
Evolutionary Computation for Modelling and Optimization in Finance
No full textIn the last decades, there has been a tendency to move away frommathematically tractable, but simplistic models towards more sophisticated andreal-world applicable models in nance. However, the consequence of the improvedsophistication is that the model specication and analysis is no longer mathematically tractable. Instead solutions need to be numerically approximated. For thistask, evolutionary computation heuristics are the appropriate means, because theydo not require any rigid mathematical properties of the model, such as linearityor convexity. Evolutionary algorithms are search heuristics, usually ispired by Darwinian evolution and Mendelian inheritance, which aim to determine the optimalsolution to a given problem by competition and alteration of candidate solutions ofa population. In this work, we focus on credit risk modelling and nancial portfoliooptimization to point out how evolutionary algorithms can easily provide realiableand accurate solutions to challenging financial problems
Genetic Algorithms in Partitional Clustering: a comparison
No full textThree approaches to partitional clustering using genetic algorithms (GA) are compared with k-means and the EM algorithm for three real world datasets (Iris, Glass and Vowel). The GA techniques differ in their encoding of the clustering problem using either a class id for each object (GAIE), medoids to assign objects to the class associated with the nearest medoid (GAME), or parameters for multivariate distributions that describe each cluster (GAPE). For the simple Iris dataset, all algorithms except GAIE obtained results with comparable accuracy, but k-means and EM had more runs with inferior results compared to GAME and GAPE. For the more complex Glass dataset, the results for GAME and GAPE were superior compared to k-means, EM and GAIE regarding their accuracy and variance of the results for repeated runs. None of the algorithms was superior for the most complex dataset (Vowel). We conclude that GAs in clustering are a valuable alternative to k-means and EM, but that the choice of the problem representation is crucial
- …
