David A. Van Veldhuizen and Gary B. Lamont. Multiobjective evolutionary algorithm research: A history and analysis. Technical Report TR-98-03, Air Force Institute of Technology, Wright Paterson AFB, Oct. 1998. 64  Home/Search   Document Not in Database   Summary   Related Articles   Check

....individuals per hypercube) is either increased or decreased in consequence. The constraint handling approach of the original micro GA was kept intact (see [3] for details) 4 Metrics Adopted In order to give a numerical comparison of our approach, we adopted three metrics: generational distance [15], error ratio [14] and spacing [12] The description and mathematical representation of each metric are shown below. 1. Generational Distance (GD) The concept of generational distance was introduced by Van Veldhuizen Lamont [15] as a way of estimating how far are the elements in the set of ....

....of our approach, we adopted three metrics: generational distance [15] error ratio [14] and spacing [12] The description and mathematical representation of each metric are shown below. 1. Generational Distance (GD) The concept of generational distance was introduced by Van Veldhuizen Lamont [15] as a way of estimating how far are the elements in the set of nondominated vectors found so far from those in the Pareto optimal set and is defined as: 329 (1) is the number of vectors in the set of nondominated solutions found so far and is the Euclidean distance ....

David A. Van Veldhuizen and Gary B. Lamont. Multiobjective Evolutionary Algorithm Research: A History and Analysis. Technical Report TR-98-03, Department of Electrical and Computer Engineering, Graduate School of Engineering, Air Force Institute of Technology, Wright-Patterson AFB, Ohio, 1998.

....applications . In both cases, in multi objective optimisation we have a multi objective function to optimise under a set of constraints: max(f(x) fk(x) s.t. g(x,p) 0, gt(x,p) 0 (1) This problem is well known and a number of approaches, both non genetic [1] and genetic algorithm [2] approaches exist. An additional problem is that not all objectives are equally important which necessitates the use of weights or preferences. We have applied appropriate genetic algorithms for multi objective optimisation and described a number of the design problems in [3] According to the ....

David A. Van Veldhuizen and Gary B. Lamont. Multiobjective evolutionary algorithm research: A history and analysis. Technical Report TR-98-03, Air Force Institute of Technology, Wright Paterson AFB, October 1998.

....multimodal problems into very dicult multiobjective optimization problems. More recently, his proposal was extended to constrained multiobjective optimization problems [17] in most of the early papers on EMOO techniques, only unconstrained test functions were used) Van Veldhuizen and Lamont [66, 67] have also proposed some guidelines to design a test function suite for evolutionary multiobjective optimization techniques, and have included in a technical report some sample test problems (mainly combinatorial optimization problems) 66] In this regard, the literature on multiobjective ....

.... test functions were used) Van Veldhuizen and Lamont [66, 67] have also proposed some guidelines to design a test function suite for evolutionary multiobjective optimization techniques, and have included in a technical report some sample test problems (mainly combinatorial optimization problems) [66]. In this regard, the literature on multiobjective combinatorial optimization can be quite useful [23] The benchmarks available for problems like the multiobjective 0 1 knapsack can be used to validate EMOO approaches. Such idea has been explored by a few EMOO researchers (for example [78, 39] ....

David A. Van Veldhuizen and Gary B. Lamont. Multiobjective Evolutionary Algorithm Research: A History and Analysis. Technical Report TR-98-03, Department of Electrical and Computer Engineering, Graduate School of Engineering, Air Force Institute of Technology, Wright-Patterson AFB, Ohio, 1998.

....these objectives are usually con icting, no single solution may exist that is best regarding all criteria considered. Therefore, at some stage of the problem solving process, the decision maker (DM) has to articulate his her preferences about the objectives. Following a classi cation by Veldhuizen [7], the articulation of preferences may be done either before (a priori) during (progressive) or after (a posteriori) the optimization process. A priori approaches are usually not practicable, since they require the user to explicitly and exactly weigh the di erent criteria and to e ectively turn ....

D. A. Van Veldhuizen and G. B. Lamont. Multiobjective evolutionary algorithm research: A history and analysis. Technical Report TR-98-03, Department of Electrical and Computer Engineering, Air Force Institute of Technology, Wright-Patterson AFB, Ohio, 1998. 7

....which make then less prone to local optimum entrapment and allow for several types of parallel implementations. The first application of GAs to MOPs dates back to the mid eighties[1, 2] and numerous papers on the application of evolutionary techniques to MOPs have been published since then (see [3, 4, 5, 6]) Although other techniques have been used for finding a suitable solution to MOPs we are interested here in those that try to approximate the Pareto set. The most popular techniques seem to be MOGA, due to Fonseca and Fleming[7] NPGA, due to Horn and Nafpliotis[8] and NSGA, due to Srinivas and ....

D.A. Van Veldhuizen and G.B. Lamont. Multiobjective Evolutionary Algorithm Research: A History and Analysis. Technical Report TR-98-03, Department of Electrical and Computer Engineering, Graduate School of Engineering, Air Force Institute of Technology, WrightPatterson AFB, Ohio, 1998.

.... Miller, 1998; Weile et al. 1996) A number of studies have also concentrated on developing new GA implementations (Kursawe, 1990; Laumanns et al. 1998; Zitzler and Thiele, 1998) Fonseca and Fleming (1995) and Horn (1997) presented overviews of different multiobjective GA implementations, and Van Veldhuizen and Lamont (1998) made a survey of test problems that exist in the literature. Despite these interests, there seems to be a lack of studies discussing problem features that may cause difficulty for multi objective GAs. The literature also lacks a set of c #1999 by the Massachusetts Institute of Technology ....

Van Veldhuizen, D. and Lamont, G. B. (1998). Multiobjective evolutionary algorithm research: A history and analysis. Technical Report Number TR-98-03. Wright-Patterson AFB, Department of Electrical and Computer Engineering, Air Force Institute of Technology, Ohio.

....simply because in most test problems tried so far most of the search algorithms based on non domination concept has found the true Pareto optimal front. With the development of many new algorithms, many researchers have attempted to summarize the studies in the field from different perspectives [5, 13, 19, 34, 2]. These reviews list many different techniques of multi criterion optimization that exist to date. A web site maintained by Carlos A. Coello Coello (http: www.lania.mx ccoello EMOO EMOObib.html) shows that there are at least 300 research papers written till to date. A year wise count of those ....

van Veldhuizen, D. and Lamont, G. B. (1998). Multiobjective evolutionary algorithm research: A history and analysis. Report Number TR-98-03. Wright-Patterson AFB, Ohio: Department of Electrical and Computer Engineering, Air Force Institute of Technology.

....devise. An example of this approach is a sum of weights of the form: min k X i=1 w i f i ( x) 6) where w i 0 are the weighting coefficients representing the relative importance of the objectives. It is usually assumed that k X i=1 w i = 1 (7) 2 The interested reader should refer to [4, 5] for more detailed surveys of EMOO approaches. 4.1.1 Applications Syswerda and Palmucci [6] used weights in their fitness function to add or subtract values during the schedule evaluation of a resource scheduler, depending on the existence or absence of penalties (constraints violated) Jakob et ....

....(computationally speaking) because they do not require any nondominance comparisons. However, their main weakness is the definition of the goals (and probably weights for each objective) These techniques have also been criticized for not being able to deal with non convex search spaces [5]. 5 Theory Not much theoretical work has been performed in this area, despite the large amount of publications reported in the literature, since most of them deal with either applications or new variations of existing techniques. The most important theoretical work in this area is easily ....

[Article contains additional citation context not shown here]

David A. Van Veldhuizen and Gary B. Lamont. Multiobjective Evolutionary Algorithm Research: A History and Analysis. Technical Report TR-98-03, Department of Electrical and Computer Engineering, Graduate School of Engineering, Air Force Institute of Technology, Wright-Patterson AFB, Ohio, 1998.

....simply because in most test problems tried so far most of the search algorithms based on non domination concept has found the true Pareto optimal front. With the development of many new algorithms, many researchers have attempted to summarize the studies in the field from different perspectives [18, 25, 42, 3]. These reviews list many different techniques of multi criterion optimization that exist to date. We now present a brief summary of a few salient evolutionary multicriterion optimization algorithms. 1.4.1 Schaffer s VEGA Schaffer [36] modified the simple tripartite genetic algorithm by ....

....on other issues. With the development of multi objective optimization algorithms, there is an obvious need of a good set of test problems, which will test and compare different multi criterion optimization algorithms with each other. Many test problems that are used in the literature are listed in [42]. However, it is not clear what aspect of an algorithm is tested by these test problems. Recently, a systematic procedure of constructing test problems have been suggested [7] In that study, test problems are constructed from single objective optimization problems to test two main features a ....

van Veldhuizen, D. and Lamont, G. B. (1998). Multiobjective evolutionary algorithm research: A history and analysis. Report Number TR-98-03. Wright-Patterson AFB, Ohio: Department of Electrical and Computer Engineering, Air Force Institute of Technology.

....can state it as follows: Find the vector x = x 1 ; x 2 ; x n ] T which will satisfy the m inequality constraints: g i ( x) 0 i = 1; 2; m (1) 2 Right after the submission of this paper, David A. Van Veldhuizen and Gary B. Lamont made available a technical report [99] that contains another remarkable survey of the area that complements the material contained in this paper. the p equality constraints h i ( x) 0 i = 1; 2; p (2) and optimizes the vector function f( x) f 1 ( x) f 2 ( x) f k ( x) T (3) where x = x 1 ; x 2 ; ....

....massively multimodal problems into very difficult multiobjective optimization problems. However, in his technical report, Deb [14] only defines test problems with 2 objective functions and the scalability of these test functions to more objectives is not straightforward. Van Veldhuizen and Lamont [99] have also proposed some guidelines to design a test function suite for evolutionary multiobjective optimization techniques, and have included in their report some sample test problems. Using benchmark problems such as those proposed by Deb [14] and Van Veldhuizen Lamont [99] it should be ....

[Article contains additional citation context not shown here]

David A. Van Veldhuizen and Gary B. Lamont. Multiobjective Evolutionary Algorithm Research: A History and Analysis. Technical Report TR-98-03, Department of Electrical and Computer Engineering, Graduate School of Engineering, Air Force Institute of Technology, Wright-Patterson AFB, Ohio, 1998.

.... They were born in 1985 when Schaffer [16] and Fourman [6] implemented the first MOEAs dealing with Multiobjective Optimization Problems (MOPs) Since then, over 140 published papers propose various MOEA implementations and applications, and to a much lesser extent, underlying MOEA theory [19]. Many of these efforts use numeric MOPs as examples to show algorithmic performance. Nowhere in the literature, however, is there a comprehensive discussion of MOP landscape issues; nor is there any explanation of why numeric MOPs may be appropriate MOEA test functions. To date, most MOEA ....

....test functions. To date, most MOEA researchers modus operandi is an algorithm s comparison (usually the researcher s own new and improved variant) against an older MOEA and analyzing results for some MOP (typically Schaffer s MOEA and examples [16] Many other example numeric MOPs are also used [19]. Results are often clearly shown in graphical form, indicating which algorithm is more effective. These empirical comparisons do not contribute much to a common basis for MOEA comparison. The literature s history of visually comparing MOEA performance on nonstandard and unjustified numeric ....

[Article contains additional citation context not shown here]

Van Veldhuizen, David A. and Gary B. Lamont. Multiobjective Evolutionary Algorithm Research: A History and Analysis. Technical Report TR-98-03, Air Force Institute of Technology, 1998.

....of visually comparing MOEA performance on non standard and unjustified numeric MOPs does little to determine a given MOEA s actual efficiency and effectiveness. A standard suite of numeric functions exhibiting relevant MOP domain characteristics can provide the necessary common comparative basis [2]. The MOEA community s limited de facto test suite contains various functions, many of whose origins and rationale for use are unknown. 1, 2] The lack of complex mathematical MOEA performance assessment tests implies that identification of appropriate functions to objectively determine MOEA ....

....and effectiveness. A standard suite of numeric functions exhibiting relevant MOP domain characteristics can provide the necessary common comparative basis [2] The MOEA community s limited de facto test suite contains various functions, many of whose origins and rationale for use are unknown. [1, 2]. The lack of complex mathematical MOEA performance assessment tests implies that identification of appropriate functions to objectively determine MOEA efficiency and effectiveness is required. Thus, a documented MOP test suite is an asset to MOEA research. We provide various MOPs for use in a ....

[Article contains additional citation context not shown here]

Van Veldhuizen, David A. and Gary B. Lamont. Multiobjective Evolutionary Algorithm Research: A History and Analysis . Technical Report TR98 -03, Air Force Institute of Technology, 1998.

.... with Multiobjective Evolutionary Algorithms (MOEAs) presents several unique challenges (e.g. vector valued fitness, sets of solutions, evolutionary operator complexity) For a good introduction to the relevant issues and past EA based approaches, see the articles by Van Veldhuizen and Lamont [5], Coello [1] and by Fonseca and Fleming [2] This paper focuses on the traditional notion of building blocks, extending the concept to the MOP domain in an effort to develop more effective and efficient MOEAs. This is a new, innovative MOEA approach. 2 MOPs and Building Blocks When considered ....

Van Veldhuizen, David A. and Gary B. Lamont. Multiobjective Evolutionary Algorithm Research: A History and Analysis . Technical Report TR98 -03, Air Force Institute of Technology, 1998.

No context found.

David A. Van Veldhuizen and Gary B. Lamont. Multiobjective evolutionary algorithm research: A history and analysis. Technical Report TR-98-03, Air Force Institute of Technology, Wright Paterson AFB, Oct. 1998. 64

No context found.

Van Veldhuizen, D. A. and Lamont, G. B. (1998b). Multiobjective evolutionary algorithm research: A history and analysis. Technical Report TR-98-03, Department of Electrical and Computer Engineering, Graduate School of Engineering, Air Force Institute of Technology, WrightPatterson AFB, Ohio.

No context found.

Van Veldhuizen D.A. and Lamont G.B. (1999). Multiobjective Evolutionary Algorithm Research: A History and Analysis, Ph.D. Thesis.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC