Multi-objective Evolutionary Algorithms in Real-World Applications: Some Recent Results and Current Challenges

Coello, Carlos A. Coello

doi:10.1007/978-3-319-11541-2_1

Cited by 31 publications

(8 citation statements)

References 73 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The most common and basic approaches to tackle multi‐objective problems using evolutionary computation consider all but one objective as constraints or the combination of the individual objective functions into a single aggregative function [24]. Other more powerful approaches attempt to determine a Pareto‐optimal or non‐dominated set of solutions [24]. This means a set of candidate solutions offering different objective trade‐offs, and for which none of the objectives can be improved without detriment of other objective function.…”

Section: Methodsmentioning

confidence: 99%

“…The MOGA differs from the classical GA only in the way fitness is obtained for each individual in the population. A rank is first assigned to each solution, according to the number of chromosomes in the population by which it is dominated [24]. Then, a fitness is assigned to every solution based on its rank [30].…”

Section: Gas With Multiple Objectivesmentioning

confidence: 99%

“…This wrapper was focused only on classification accuracy improvement, overlooking other important issues such as the dimensionality of the representation. To obtain a more proper representation for speech recognition, here we propose a multi‐objective GA (MOGA) [24], which allows to maximise the classification accuracy while minimising the number of features. In this case, for the purpose of obtaining appropriate features for state‐of‐the‐art speech recognisers, a classifier based on hidden Markov models (HMMs) [25] is used to estimate the capability of candidate solutions, using on a set of English phonemes.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Multi‐objective optimisation of wavelet features for phoneme recognition

Vignolo

Rufiner

Milone

2016

IET signal process.

View full text Add to dashboard Cite

One of the most important issues in speech applications involves the preprocessing stage, which is meant to produce a manageable set of significant features, exploiting the capabilities of the classification phase [5]. The most widely used features for speech recognition, and also applied for different tasks involving speech and music signals, are the mel-frequency cepstral coefficients (MFCCs) [1,5]. These are based on the linear model of voice production and a psychoacoustic scale [5]. Even though MFCCs provide acceptable performance under laboratory conditions, recognition rates degrade significantly in presence of noise. This has motivated many advances in the development of robust feature extraction approaches, like perceptual linear prediction (PLP) and relative spectra [1]. More recently, speech processing techniques based on computational intelligence tools have been developed. For example, several approaches based on evolutionary computation have been proposed for the search of optimal speech representations [8]. Wavelet based processing provides useful tools for the analysis of nonstationary signals, which have been found suitable for speech feature extraction [6]. In order to build a representation based on the wavelet packet transform (WPT), frequently a particular orthogonal basis is selected among all the available basis [6]. However, for speech recognition there is no evidence showing the convenience of the use of orthogonal basis. Therefore, removing the orthogonality restriction the complete WPT decomposition offers a highly redundant set of coefficients, some of which can be selected to build an optimal representation.The optimisation of wavelet decompositions for feature extraction has been studied in many different ways, though it is still an open challenge in speech processing. For example, the optimisation of wavelet decompositions by means of evolutionary algorithms was proposed for image watermarking [4] and for signal denoising [2]. In [9] we proposed a novel approach for the optimisation of over-complete decompositions from a WPT dictionary based on a multi-objective genetic algorithm (MOGA). The MOGA allows to maximise the classification accuracy while minimising the number of features. For the purpose of obtaining appropriate features for state of the art speech recognizers, a classifier based on hidden Markov models (HMM) is used to estimate the capability of candidate solutions, using on a set of English phonemes. The proposed method, which we refer to as evolutionary wavelet packets (EWP), exploits the benefits provided

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Gas With Multiple Objectivesmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Multi‐objective optimisation of wavelet features for phoneme recognition

Vignolo

Rufiner

Milone

2016

IET signal process.

View full text Add to dashboard Cite

show abstract

“…For example, the Map-Reduce paradigm adopted by Hadoop has been used for the parallelisation of single objective Genetic Algorithms [5,7,8,12], where the main gain is in the distributed evaluation of a single fitness function. When moving towards the need to optimize two or more objectives, the fitness evaluation itself is no longer the only expensive component [3]: The construction of the Pareto Front (PF) becomes computationally demanding [4].…”

Section: Introductionmentioning

confidence: 99%

A new approach to distribute MOEA pareto front computation

Sarro

Petrozziello

et al. 2020

Proceedings of the 2020 Genetic and Evolutionary Computation Conference Companion

View full text Add to dashboard Cite

Multi-Objective Evolutionary Algorithms (MOEAs) offer compelling solutions to many real world problems, including software engineering ones. However, their efficiency decreases with the growing size of the problems at hand, hindering their applicability in practice. In this paper we propose a novel master-worker approach to distribute the computation of the Pareto Front (PF) for MOEAs (dubbed MOEA-DPF) and empirically evaluate it on a real-world software project management problem. With respect to previous work, our proposal can be used with any MOEA to tackle multiobjective problems regardless of their formulation/representation. Our results show that MOEA-DPF runs significantly faster (up to 3.1x speed-up using two workers) than its sequential counterpart while maintaining (and even improving) the quality of the PF. We conclude that MOEA-DPF provides an effective and simple solution to speed-up the execution of MOEAs by distributing the PF computation, making them effective for real-world problems. CCS CONCEPTS • Software and its engineering → Search-based software engineering; • Computing methodologies → Distributed algorithms;

show abstract

“…Different from single-objective optimization, the task of multiobjective optimization is to achieve a set of optimal nondominated solutions uniformly distributed in the objective space. Due to the inherent ability of evolving a swarm of solutions simultaneously at a generation, evolution algorithms have been widely adopted in multiobjective optimization [20,21]. Among the multiobjective evolutionary algorithms (MOEAs), multiobjective differential evolution (MODE) has achieved a lot of successful applications, especially in practical industries, due to its simple but effective search mechanism [22][23][24][25][26][27][28][29].…”

Section: Brief Introduction Of Multiobjective Differentialmentioning

confidence: 99%

Integrated Scheduling of Reheating Furnace and Hot Rolling Based on Improved Multiobjective Differential Evolution

Tian

2018

Complexity

View full text Add to dashboard Cite

In the hot rolling mill, a cold slab is reheated in reheating furnace and then rolled into thin coils through hot rolling process. Traditional research on hot rolling mill generally takes the two processes as two separate single-objective scheduling problems, though they are closely connected and their optimization objectives are conflicting with each other. In this paper, the integrated scheduling of the reheating furnace and hot rolling is investigated in the view of multiobjective optimization. A mixed-integer programming model with two objectives is formulated for this problem, and a multiobjective differential evolution (MODE) algorithm is developed to solve this model. To make it easy for algorithm design, a hot rolling schedule (i.e., a sequence of slabs) is used to represent a solution and for a given solution, a heuristic with consideration of energy consumption minimization is presented to obtain the schedule of reheating furnace. To achieve the balance of exploration and exploitation, a novel mutation operator with adaptive leader selection, a crossover operator with feasibility consideration, and a guided multiobjective local search are designed. Computational results based on simulated data illustrate that the integrated scheduling is superior to the traditional two-phase scheduling method used in practice and literature. Further results illustrate that the proposed MODE algorithm is effective and superior to some other multiobjective evolutionary algorithms.

show abstract

Multi-objective Evolutionary Algorithms in Real-World Applications: Some Recent Results and Current Challenges

Cited by 31 publications

References 73 publications

Multi‐objective optimisation of wavelet features for phoneme recognition

Multi‐objective optimisation of wavelet features for phoneme recognition

A new approach to distribute MOEA pareto front computation

Integrated Scheduling of Reheating Furnace and Hot Rolling Based on Improved Multiobjective Differential Evolution

Contact Info

Product

Resources

About