Karthik Sindhya

A Hybrid Framework for Evolutionary Multi-Objective Optimization

Evolutionary multi-objective optimization algorithms are widely used for solving optimization problems with multiple conflicting objectives. However, basic evolutionary multi-objective optimization algorithms have shortcomings, such as slow convergence to the Pareto optimal front, no efficient termination criterion, and a lack of a theoretical convergence proof. A hybrid evolutionary multi-objective optimization algorithm involving a local search module is often used to overcome these shortcomings. But there are many issues that affect the performance of hybrid evolutionary multi-objective optimization algorithms, such as the type of scalarization function used in a local search and frequency of a local search. In this paper, we address some of these issues and propose a hybrid evolutionary multi-objective optimization framework. The proposed hybrid evolutionary multi-objective optimization framework has a modular structure, which can be used for implementing a hybrid evolutionary multi-objective optimization algorithm. A sample implementation of this framework considering NSGA-II, MOEA/D, and MOEA/D-DRA as evolutionary multi-objective optimization algorithms is presented. A gradient-based sequential quadratic programming method as a single objective optimization method for solving a scalarizing function used in a local search is implemented. Hence, only continuously differentiable functions were considered for numerical experiments. The numerical experiments demonstrate the usefulness of our proposed framework.

A preference based interactive evolutionary algorithm for multi-objective optimization: PIE

This paper describes a new Preference-based Interactive Evolutionary (PIE) algorithm for multi-objective optimization which exploits the advantages of both evolutionary algorithms and multiple criteria decision making approaches. Our algorithm uses achievement scalarizing functions and the potential of population based evolutionary algorithms to help the decision maker to direct the search towards the desired Pareto optimal solution. Starting from an approximated nadir point, the PIE algorithm improves progressively the objective function values of a solution by finding a better solution at each iteration that improves the previous one. The decision maker decides from which solution, in which direction, and at what distance from the Pareto front to find the next solution. Thus, the PIE algorithm is guided interactively by the decision maker. A flexible approach is obtained with the use of archive sets to store all the solutions generated during an evolutionary algorithms run, as it allows the decision maker to freely navigate and inspect previous solutions if needed. The PIE algorithm is demonstrated using a pollution monitoring station problem and shown to be effective in helping the decision maker to find a solution that satisfies her/his preferences.

Local search based evolutionary multi-objective optimization algorithm for constrained and unconstrained problems

Evolutionary multi-objective optimization algorithms are commonly used to obtain a set of non-dominated solutions for over a decade. Recently, a lot of emphasis have been laid on hybridizing evolutionary algorithms with MCDM and mathematical programming algorithms to yield a computationally efficient and convergent procedure. In this paper, we test an augmented local search based EMO procedure rigorously on a test suite of constrained and unconstrained multi-objective optimization problems. The success of our approach on most of the test problems not only provides confidence but also stresses the importance of hybrid evolutionary algorithms in solving multi-objective optimization problems.

Self-adaptive simulated binary crossover for real-parameter optimization

Simulated binary crossover (SBX) is a real-parameter recombinationoperator which is commonly used in the evolutionary algorithm (EA) literature. The operatorinvolves a parameter which dictates the spread of offspring solutionsvis-a-vis that of the parent solutions. In all applications of SBX sofar, researchers have kept a fixed value throughout a simulation run. In this paper, we suggest a self-adaptive procedure of updating theparameter so as to allow a smooth navigation over the functionlandscape with iteration. Some basic principles of classicaloptimization literature are utilized for this purpose. The resultingEAs are found to produce remarkable and much better results comparedto the original operator having a fixed value of the parameter. Studieson both single and multiple objective optimization problems are madewith success.

A Local Search Based Evolutionary Multi-objective Optimization Approach for Fast and Accurate Convergence

A local search method is often introduced in an evolutionary optimization technique to enhance its speed and accuracy of convergence to true optimal solutions. In multi-objective optimization problems, the implementation of a local search is a non-trivial task, as determining a goal for the local search in presence of multiple conflicting objectives becomes a difficult proposition. In this paper, we borrow a multiple criteria decision making concept of employing a reference point based approach of minimizing an achievement scalarizing function and include it as a search operator of an EMO algorithm. Simulation results with NSGA-II on a number of two to four-objective problems with and without the local search approach clearly show the importance of local search in aiding a computationally faster and more accurate convergence to Pareto-optimal solutions. The concept is now ready to be coupled with a faster and more accurate diversity-preserving procedure to make the overall procedure a competitive algorithm for multi-objective optimization.

A new hybrid mutation operator for multiobjective optimization with differential evolution

Differential evolution has become one of the most widely used evolutionary algorithms in multiobjective optimization. Its linear mutation operator is a simple and powerful mechanism to generate trial vectors. However, the performance of the mutation operator can be improved by including a nonlinear part. In this paper, we propose a new hybrid mutation operator consisting of a polynomial-based operator with nonlinear curve tracking capabilities and the differential evolution’s original mutation operator, for the efficient handling of various interdependencies between decision variables. The resulting hybrid operator is straightforward to implement and can be used within most evolutionary algorithms. Particularly, it can be used as a replacement in all algorithms utilizing the original mutation operator of differential evolution. We demonstrate how the new hybrid operator can be used by incorporating it into MOEA/D, a winning evolutionary multiobjective algorithm in a recent competition. The usefulness of the hybrid operator is demonstrated with extensive numerical experiments showing improvements in performance compared with the previous state of the art.

Hybridization of SBX based NSGA-II and sequential quadratic programming for solving multi-objective optimization problems

Most real-world search and optimization problems involve multiple conflicting objectives and results in a Pareto-optimal set. Various multi-objective optimization algorithms have been proposed for solving such problems with the goals of finding as many trade-off solutions as possible and maintaining diversity among them. Since last decade, evolutionary multi-objective optimization (EMO) algorithms have been applied successfully to various test and real-world optimization problems. These population based algorithms provide a diverse set of non-dominated solutions. The obtained non-dominated set is close to the true Pareto-optimal front but its convergence to the true Pareto-optimal front is not guaranteed. Hence to ensure the same, a local search method using classical algorithm can be applied. In the present work, SBX based NSGA-II is used as a population based approach and the sequential quadratic programming (SQP) method is used as a local search procedure. This hybridization of evolutionary and classical algorithms approach provides a confidence of converging near to the true Pareto-optimal set with a good diversity. The proposed procedure is successfully applied to 13 test problems consisting two, three and five objectives. The obtained results validate our motivation of hybridizing evolutionary and classical methods.

Improving convergence of evolutionary multi-objective optimization with local search: a concurrent-hybrid algorithm

A local search method is often introduced in an evolutionary optimization algorithm, to enhance its speed and accuracy of convergence to optimal solutions. In multi-objective optimization problems, the implementation of local search is a non-trivial task, as determining a goal for local search in presence of multiple conflicting objectives becomes a difficult task. In this paper, we borrow a multiple criteria decision making concept of employing a reference point based approach of minimizing an achievement scalarizing function and integrate it as a search operator with a concurrent approach in an evolutionary multi-objective algorithm. Simulation results of the new concurrent-hybrid algorithm on several two to four-objective problems compared to a serial approach, clearly show the importance of local search in aiding a computationally faster and accurate convergence to the Pareto optimal front.

A Surrogate-Assisted Reference Vector Guided Evolutionary Algorithm for Computationally Expensive Many-Objective Optimization

We propose a surrogate-assisted reference vector guided evolutionary algorithm (EA) for computationally expensive optimization problems with more than three objectives. The proposed algorithm is based on a recently developed EA for many-objective optimization that relies on a set of adaptive reference vectors for selection. The proposed surrogate-assisted EA (SAEA) uses Kriging to approximate each objective function to reduce the computational cost. In managing the Kriging models, the algorithm focuses on the balance of diversity and convergence by making use of the uncertainty information in the approximated objective values given by the Kriging models, the distribution of the reference vectors as well as the location of the individuals. In addition, we design a strategy for choosing data for training the Kriging model to limit the computation time without impairing the approximation accuracy. Empirical results on comparing the new algorithm with the state-of-the-art SAEAs on a number of benchmark problems demonstrate the competitiveness of the proposed algorithm.

An Interactive Simple Indicator-Based Evolutionary Algorithm (I-SIBEA) for Multiobjective Optimization Problems

This paper presents a new preference based interactive evolutionary algorithm (I-SIBEA) for solving multiobjective optimization problems using weighted hypervolume. Here the decision maker iteratively provides her/his preference information in the form of identifying preferred and/or non-preferred solutions from a set of nondominated solutions. This preference information provided by the decision maker is used to assign weights of the weighted hypervolume calculation to solutions in subsequent generations. In any generation, the weighted hypervolume is calculated and solutions are selected to the next generation based on their contribution to the weighted hypervolume. The algorithm is compared with a recently developed interactive evolutionary algorithm, W-Hype on some benchmark multiobjective optimization problems. The results show significant promise in the use of the I-SIBEA algorithm. In addition, the performance of the algorithm is demonstrated using a human decision maker to show its flexibility towards changes in the preference information. The I-SIBEA algorithm is found to flexibly exploit the preference information from the decision maker and generate solutions in the regions preferable to her/him.