A practical tutorial on the use of nonparametric statistical tests as a methodology for comparing evolutionary and swarm intelligence algorithms

Derrac, Joaquín; García, Salvador; Molina, Daniel; Herrera, Francisco

doi:10.1016/j.swevo.2011.02.002

Cited by 4,375 publications

(1,850 citation statements)

References 29 publications

Supporting

Mentioning

1,635

Contrasting

Unclassified

Order By: Relevance

“…In addition, we hope to show the performance of DCABC by Null Hypothesis Significance Testing (NHST) [35,36] in our future work. We only test the new algorithm on classical benchmark functions and have not used it to solve practical problems, such as fault diagnosis [37], path plan [38], Knapsack [39][40][41], multi-objective optimization [42], gesture segmentation [43], unit commitment problem [44], and so on.…”

Section: Resultsmentioning

confidence: 99%

A Developed Artificial Bee Colony Algorithm Based on Cloud Model

Jin

Sun

2018

Mathematics

View full text Add to dashboard Cite

Abstract:The Artificial Bee Colony (ABC) algorithm is a bionic intelligent optimization method. The cloud model is a kind of uncertainty conversion model between a qualitative concept T that is presented by nature language and its quantitative expression, which integrates probability theory and the fuzzy mathematics. A developed ABC algorithm based on cloud model is proposed to enhance accuracy of the basic ABC algorithm and avoid getting trapped into local optima by introducing a new select mechanism, replacing the onlooker bees' search formula and changing the scout bees' updating formula. Experiments on CEC15 show that the new algorithm has a faster convergence speed and higher accuracy than the basic ABC and some cloud model based ABC variants.

show abstract

Section: Resultsmentioning

confidence: 99%

A Developed Artificial Bee Colony Algorithm Based on Cloud Model

Jin

Sun

2018

Mathematics

View full text Add to dashboard Cite

show abstract

“…To verify this hypothesis, statistical analysis was performed. Table 4 reports average ranking of the methods after the Friedman test (Derrac et al, 2011). For each configuration, the lowest rank is marked in bold.…”

Section: Comparison Of the Algorithmsmentioning

confidence: 99%

“…For each configuration, the lowest rank is marked in bold. Next, statistically important (based on the Shaffer test with α = 0.05 level of significance (Derrac et al, 2011) and ranking after the Friedman test). According to the results, the CG rank is always the smallest and very close to 1 while the differences between CG results and results of any other non-exact method are statistically important.…”

Section: Comparison Of the Algorithmsmentioning

confidence: 99%

A column generation technique for routing and spectrum allocation in cloud–ready survivable elastic optical networks

Goścień

Walkowiak

2017

International Journal of Applied Mathematics and Computer Science

View full text Add to dashboard Cite

Driven by increasing user requirements and expectations, the fast development of telecommunications networks brings new challenging optimization problems. One of them is routing and spectrum allocation (RSA) of three types of network flows (unicast, anycast, multicast) in elastic optical networks (EONs) implementing dedicated path protection (DPP). In the paper, we model this problem as integer linear programming (ILP) and we introduce two new optimization approaches-a dedicated heuristic algorithm and a column generation (CG)-based method. Then, relying on extensive simulations, we compare algorithm performance with reference methods and evaluate CG efficiency in detail. The results show that the proposed CG method significantly outperforms reference algorithms and achieves results very close to optimal ones (the average distance to optimal results was at most 2.1%).

show abstract

“…In order to comply with reported best practice in the evaluation of the performance of neural networks, (Luengo et al, 2009;García et al, 2010;Derrac et al, 2011), we evaluated the statistical significance of the observed performance results applying the Friedman test. This test ranks the performance of a set of k algorithms and can detect a significant difference in the performance of at least two algorithms.…”

Section: Non-parametric Statistical Analysis and Posthoc Proceduresmentioning

confidence: 99%

“…For the sake of our evaluation we need to carry out a multiple comparisons analysis between performance of the LIT-Approach and performance of each one of the other initialization methods. This is a multiple comparisons (pairwise) analysis (Derrac et al, 2011) with a control algorithm which results in formulating k − 1 hypotheses one for each of the k − 1 comparisons, where in our case k = 7. A better performance for the convergence rate of an algorithm trans- lates here to a smaller number of epochs and a better performance for generalization is taken to be a smaller classification or approximation error.…”

Section: Non-parametric Statistical Analysis and Posthoc Proceduresmentioning

confidence: 99%

Solving the linear interval tolerance problem for weight initialization of neural networks

Adam¹,

Karras²,

Magoulas³

et al. 2014

Neural Networks

View full text Add to dashboard Cite

Determining good initial conditions for an algorithm used to train a neural network is considered a parameter estimation problem dealing with uncertainty about the initial weights. Interval Analysis approaches model uncertainty in parameter estimation problems using intervals and formulating tolerance problems. Solving a tolerance problem is defining lower and upper bounds of the intervals so that the system functionality is guaranteed within predefined limits. The aim of this paper is to show how the problem of determining the initial weight intervals of a neural network can be defined in terms of solving a linear interval tolerance problem. The proposed Linear Interval Tolerance Approach copes with uncertainty about the initial weights without any previous knowledge or specific assumptions on the input data as required by approaches such as fuzzy sets or rough sets. The proposed method is tested on a number of well known benchmarks for neural networks trained with the back-propagation family of algorithms. Its efficiency is evaluated with regards to standard performance measures and the results obtained are compared against results of a number of well known and established initialization methods. These results provide credible evidence that the proposed method outperforms classical weight initialization methods.

show abstract

A practical tutorial on the use of nonparametric statistical tests as a methodology for comparing evolutionary and swarm intelligence algorithms

Cited by 4,375 publications

References 29 publications

A Developed Artificial Bee Colony Algorithm Based on Cloud Model

A Developed Artificial Bee Colony Algorithm Based on Cloud Model

A column generation technique for routing and spectrum allocation in cloud–ready survivable elastic optical networks

Solving the linear interval tolerance problem for weight initialization of neural networks

Contact Info

Product

Resources

About