Parameter tuning of PBIL and CHC evolutionary algorithms applied to solve the Root Identification Problem
Introduction
Modern computer aided design and manufacturing systems are built on top of parametric geometric modeling engines. The field has developed sketching systems that automatically instantiate geometric objects from a rough sketch, annotated with dimensions and constraints input by the user. The sketch has only to be topologically correct and constraints are usually not yet satisfied.
Geometric problems defined by constraints have an exponential number of solution instances in the number of geometric elements involved. Generally, the user is only interested in one solution instance such that besides fulfilling the geometric constraints, it exhibits some additional properties. This solution instance is called the intended solution.
Selecting a solution instance amounts to selecting one among a number of different roots of a nonlinear equation or system of equations. The problem of selecting a given root was named the Root Identification Problem [1].
Several approaches to solve the Root Identification Problem have been reported in the literature. Examples are: selectively moving the geometric elements, conducting a dialogue with the constraint solver that identifies interactively the intended solution, and preserving the topology of the sketch input by the user. For a discussion of these approaches see, for example [1], [2] and references therein. Not only it is important to have in mind the interactive nature of computer aided design and manufacturing systems [3], [4], but also the high difficulty of solving the problem [5].
A new technique to automatically solve the Root Identification Problem for constructive solvers [6], was reported in [2], [7], [8]. The technique overconstrains the geometric problem by defining two different categories of constraints. One category includes the set of constraints specifically needed to solve the geometric constraint problem. The other category includes a set of extra constraints or predicates on the geometric elements which identify the intended solution instance.
Once the constructive solver has generated the space of solution instances, the extra constraints are used to drive an automatic search of the solution instances space using genetic algorithms [9]. The search outputs a solution instance that maximizes the number of extra constraints fulfilled.
Genetic algorithms are a category of evolutionary algorithms. Evolutionary algorithms (or, as a matter of fact, any heuristic method) are characterized by a set of parameters that determine the evolution of the algorithms for which specific values must be chosen. One of the main difficulties of applying an evolutionary algorithm to a given problem is to decide on an appropriate set of parameter values. These values are commonly chosen in practice by trial and error, tuned by hand, or taken from other fields [10].
Unfortunately, these methods do not guarantee the obtention of optimal parameter settings and since experimental results are always problem dependent, values that proved to be useful in other applications may not perform well on untried tasks [11]. Furthermore, when parameters are chosen in these ways, the relationship between the parameter values and the performance of the algorithm cannot be established.
In a previous work, see [12], two algorithms, among a large set of different population-based and single-point metaheuristics [13], [14], have been demonstrated to yield the best results when they are applied to the Root Identification Problem: Population-Based Incremental Learning (PBIL) [15], and Cross generational elitist selection Heterogeneous recombination and Cataclismic mutation (CHC) [16].
PBIL and CHC are two evolutionary algorithms that have received a large amount of attention as general purpose function optimizers. The PBIL algorithm is a method that combines generational mechanisms with simple competitive learning. It is argued that this algorithm is simple and outperforms genetic algorithms on a large set of optimization problems. CHC is a nontraditional genetic algorithm whose crossover operation is highly disruptive. This results in a search ability more effective than that of traditional genetic algorithms by balancing diversity and convergence.
In this work, we report on an empirical statistical study conducted to establish the impact of the driving parameters in the PBIL and CHC evolutionary algorithms when they are used to solve the Root Identification Problem in the field of Geometric Constraint Solving. Then we identify a set of values that optimize algorithms performance. The driving parameters considered for the PBIL algorithm are the population size, mutation probability, mutation shift and learning rate. For the CHC algorithm we studied population size, divergence rate, differential threshold and the set of best individuals. In both cases we applied unifactorial and multifactorial analysis, post hoc tests and best parameter level selection. Experimental results show that CHC outperforms PBIL when applied to solve the Root Identification Problem.
The remainder of this work is organized as follows. Section 2 briefly describes the main concepts involved in the Root Identification Problem and how evolutionary algorithms can be applied to solve it. Section 3 briefly describes the PBIL and CHC algorithms. Section 4 describes the experimental set up. Experimental results are discussed in Sections 5 and 6. Section 7 briefly compares PBIL and CHC performances, leaving Section 8 to draw some conclusions and to suggest future work.
Section snippets
The Root Identification Problem
According to [1], the problem of selecting one solution to the Geometric Constraint Solving problem is known as the Root Identification Problem and consists of selecting one solution to a system of nonlinear equations among a potentially exponential number of solutions, that is, to select one root for each equation in the system. We first briefly describe the context where this problem arises. Then we give the criteria we use to define the solution instance model, that is, the solution of
The evolutionary algorithms studied
In a previous work [12], we conducted a preliminary study to asses the potential behavior of a number of metaheuristics applied to solve the Root Identification Problem. The study considered single-point and population-based metaheuristics according to the classification given in [14].
Among the single-point search methods, we considered: local search algorithms [25], simulated annealing [26], tabu search [27] and multistart local search [28], [29]. The population-based search methods studied
Design of the experiments
To study the behavior of PBIL and CHC algorithms as a function of the parameters we have applied an empirical methodology along with a statistical ANalysis Of VAriance (ANOVA) [34], [35], [36], of the experimental results. To avoid doing a complete factorial design of experiments [10], we have taken into account the limitations of the different experimental techniques (racing algorithms, sequential parameter optimization, etc.) as well as the statistical nature of the samples obtained from our
Results of PBIL algorithm
To assess the behaviour of the PBIL algorithm we have conducted a unifactorial analysis, multifactorial analysis, post hoc tests and best parameter level selection. First, one-way ANOVA has been used to figure out whether the effect of each individual parameter on the algorithm result is significant. Second, multiple factor ANOVA has been applied to elucidate how each possible combination of factors affect algorithms performance and the importance of that influence. Third, the best values for
Results of CHC algorithm
To assess the behavior of the CHC algorithm we have conducted the same set of experiments over the same benchmark of the PBIL algorithm: unifactorial analysis, multifactorial analysis, post hoc tests and best parameter level selection.
PBIL versus CHC
Values considered in this section are derived from multifactorial ANOVA analysis. Fig. 15 plots, for each problem in the benchmark, the smallest mean run-length that allowed PBIL and CHC algorithms to find a solution instance. We can see that, except for problem instances 4_1, 4_2, 4_3 and 5_1, 5_2, 5_3, the CHC algorithm clearly outperforms the PBIL algorithm.
Fig. 16 plots the mean run-length average yielded by the best parameters levels selected in Section 5 and Section 6 for each problem in
Conclusions and future work
The solution of the Root Identification Problem in Geometric Constraint Solving by means of evolutionary algorithms, specifically PBIL and CHC algorithms, is both feasible and effective.
As expected, the statistical study carried out shows that the influence of the values assigned to the parameters that characterize the behaviour of these algorithms is important. In general, the influence of parameters considered individually is greater than when considering combinations of them. values show
Acknowledgements
This research has been supported by FEDER and by CICYT of the Spanish Ministerio de Educación y Ciencia under project TIN2007-67474-C03-01.
References (49)
- et al.
Geometric constraint solver
Computer Aided Design
(1995) The CHC adaptive search algorithm: how to have safe search when engaging in nontraditional genetic recombination
- et al.
Skecth-based pruning of a solution space within a formal geometric constraint solver
Artificial Intelligence
(2000) - M. Luzón, Resolución de restricciones geométricas. selección de la solución deseada, Ph.D. Thesis, Dept. Informática,...
- et al.
Combining constructive and equational Geometric Constraint Solving techniques
ACM Transactions on Graphics
(1999) - S. Vila, Contribution to Geometric Constraint Solving in cooperative engineering, Ph.D. Thesis, Departament de...
- et al.
A graph-constructive approach to solving systems of geometric constraints
ACM Transactions on Graphics
(1997) - et al.
A brief on constraint solving
Computer-Aided Design and Applications
(2005) - et al.
Genetic algorithms for root multiselection in constructive Geometric Constraint Solving
Computer & Graphics
(2003) - E. Barreiro, Modelización y optimización de algoritmos genéticos para la selección de la solución deseada en resolución...
Adaptation in Natural and Artificial Systems
No free lunch theorems for optimization
IEEE Transaction on Evolutionary Computation
Metaheuristics: a bibliography
Annals of Operations Research
Mathematical Models
Self-adaptivity for constraint satisfaction: learning penalty functions
Third IEEE World Conference on Evolutionary Computation
Recombination
Searching the solution space in constructive Geometric Constraint Solving with genetic algorithms
Applied Intelligence
Genetic Algorithms in Search, Optimization Machine Learning
Genetic Algorithms + Data Structures = Evolution Programs
Cited by (15)
Automatic parameter tuning for Evolutionary Algorithms using a Bayesian Case-Based Reasoning system
2014, Applied Soft Computing JournalCitation Excerpt :In reality, there is no deterministic way to select the optimal parameter setting by means of an analytical formula. We are ideally interested in the underlying joint probability distribution of parameters’ values vs. performance [17] and if there exists a method to assess such a distribution based on experimental results. This is the main reason why in [17] 50 runs are made for each configuration.
A Short Survey on Population-Based Incremental Learning Algorithm
2019, 2019 IEEE Symposium Series on Computational Intelligence, SSCI 2019The States of Matter Search (SMS)
2018, Studies in Computational IntelligenceOptimal operation of reservoir systems with the symbiotic organisms search (SOS) algorithm
2017, Journal of HydroinformaticsParameter optimization methods based on computational intelligence techniques in context of sustainable computing
2017, Studies in Computational IntelligenceOptimal reactive power dispatch involving SMES by Cuckoo search algorithm
2016, International Conference on 21st Century Energy Needs - Materials, Systems and Applications, ICTFCEN 2016