Abstract
Many-objective problems pose various challenges in terms of decision making and search. One approach to tackle the resulting problems is the automatic reduction of the number of objectives such that the information loss is minimized. While in a previous work we have investigated the issue of omitting objectives, we here address the generalized problem of aggregating objectives using weighted sums. To this end, heuristics are presented that iteratively remove two objectives and replace them by a new objective representing an optimally weighted combination of them. As shown in the paper, the new reduction method can substantially reduce the information loss and thereby can be highly useful when analyzing trade-off sets after optimization as well as during search to reduce the computation overhead related to hypervolume-based fitness assignment.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
This is the case, e.g., in hypervolume-based evolutionary algorithms where the computation time per generation is exponential in the number of objectives (Brockhoff and :̧def :̧def Zitzler 2007a; :̧def :̧def Bringmann and Friedrich 2008). Also if the objective values have to be computed by time-consuming simulations, a lower number of objectives will speed up the evaluation time.
- 2.
For simplicity, we use the term objective and the corresponding weight vector interchangeably throughout the paper.
- 3.
More precisely, to 1 and 2 objectives for the 4-objective instances, to 1 and 3 for the 6-objective instances, and to 2 and 4 for the 8-objective instances.
- 4.
The number of decision variables has been set to 250.
- 5.
For all WFG problems, the number of decision variables has been also fixed to 250 and the number of position variables has been chosen to 168 and the number of distance variables to 82 such that it can be kept constant over all numbers of objectives.
- 6.
The δ-errors have been normalized to the objective values of each instance such that the difference between the highest and lowest objective value equals 1 for every objective.
- 7.
The \(\#\mathcal{P}\)-hardness proof of :̧def :̧def Bringmann and Friedrich (2008) implies that no exact polynomial algorithm for the hypervolume indicator exists unless \(\mathcal{P} = \mathcal{N}\mathcal{P}\).
- 8.
The hypervolume indicator or \(\mathcal{S}\)-measure of a solution set A ⊂ X is informally defined as the space that is dominated by the solutions in A which itself is dominating a reference set. Here, we use only a single reference point and refer to Beume et al. (2007) for an exact definition. The hypervolume indicator has always to be maximized.
References
Beume N, Naujoks B, Emmerich M (2007) SMS-EMOA: multiobjective selection based on dominated hypervolume. Eur J Oper Res 181:1653–1669
Bleuler S, Laumanns M, Thiele L, Zitzler E (2003) PISA – a platform and programming language independent interface for search algorithms. In: Fonseca CM, et al (eds) Conference on evolutionary multi-criterion optimization (EMO 2003). LNCS, vol 2632. Springer, Berlin, pp 494–508
Bringmann K, Friedrich T (2008) Approximating the volume of unions and intersections of high-dimensional geometric objects. In: Hong SH, Nagamochi H, Fukunaga T (eds) International symposium on algorithms and computation (ISAAC 2008). LNCS, vol 5369. Springer, Berlin, pp 436–447
Brockhoff D, Zitzler E (2007a) Improving hypervolume-based multiobjective evolutionary algorithms by using objective reduction methods. In: Congress on evolutionary computation (CEC 2007). IEEE Press, pp 2086–2093
Brockhoff D, Zitzler E (2007b) Offline and online objective reduction in evolutionary multiobjective optimization based on objective conflicts. TIK Report 269, Computer Engineering and Networks Laboratory (TIK), ETH Zurich
Brockhoff D, Zitzler E (2009) Objective reduction in evolutionary multiobjective optimization: theory and applications. Evol Comput 17(2):135–166
Brockhoff D, Friedrich T, Hebbinghaus N, Klein C, Neumann F, Zitzler E (2007) Do additional objectives make a problem harder? In: Thierens D, et al (eds) Genetic and evolutionary computation conference (GECCO 2007). ACM Press, New York, NY, USA, pp 765–772
Conover WJ (1999) Practical nonparametric statistics, 3rd edn. Wiley, New York
Deb K, Saxena D (2006) Searching for Pareto-optimal solutions through dimensionality reduction for certain large-dimensional multi-objective optimization problems. In: Congress on evolutionary computation (CEC 2006). IEEE Press, pp 3352–3360
Deb K, Thiele L, Laumanns M, Zitzler E (2005) Scalable test problems for evolutionary multi-objective optimization. In: Abraham A, Jain R, Goldberg R (eds) Evolutionary multiobjective optimization: theoretical advances and applications, chap 6. Springer, Berlin, pp 105–145
Fleming PJ, Purshouse RC, Lygoe RJ (2005) Many-objective optimization: an engineering design perspective. In: Coello Coello CA, et al (eds) Conference on evolutionary multi-criterion optimization (EMO 2005). LNCS, vol 3410. Springer, Berlin, pp 14–32
Huband S, Hingston P, Barone L, While L (2006) A review of multiobjective test problems and a scalable test problem toolkit. IEEE Trans Evol Comput 10(5):477–506
Hughes EJ (2007) Radar waveform optimization as a many-objective application benchmark. In: Conference on evolutionary multi-criterion optimization (EMO 2007). LNCS, vol 4403. Springer, Berlin, pp 700–714
Künzli S, Bleuler S, Thiele L, Zitzler E (2004) A computer engineering benchmark application for multiobjective optimizers. In: Coello Coello CA, Lamont G (eds) Applications of multi-objective evolutionary algorithms. World Scientific, Singapore, pp 269–294
Laumanns M, Thiele L, Deb K, Zitzler E (2002) Combining convergence and diversity in evolutionary multiobjective optimization. Evol Comput 10(3):263–282
Laumanns M, Thiele L, Zitzler E (2004) Running time analysis of evolutionary algorithms on a simplified multiobjective Knapsack problem. Nat Comput 3(1):37–51
López Jaimes A, Coello Coello CA, Chakraborty D (2008) Objective reduction using a feature selection technique. In: Conference on genetic and evolutionary computation (GECCO 2008). ACM, New York, USA, pp 673–680, doi http://doi.acm.org/10.1145/1389095.1389228
Purshouse RC, Fleming PJ (2003) Conflict, harmony, and independence: relationships in evolutionary multi-criterion optimisation. In: Conference on evolutionary multi-criterion optimization (EMO 2003). LNCS, vol 2632. Springer, Berlin, pp 16–30
Saxena DK, Deb K (2007) Non-linear dimensionality reduction procedures for certain large-dimensional multi-objective optimization problems: employing correntropy and a novel maximum variance unfolding. In: Conference on evolutionary multi-criterion optimization (EMO 2007). LNCS, vol 4403. Springer, Berlin, pp 772–787, doi 10.1007/ 978-3-540-70928-2_58
Sülflow A, Drechsler N, Drechsler R (2007) Robust multi-objective optimization in high dimensional spaces. In: Conference on evolutionary multi-criterion optimization (EMO 2007). LNCS, vol 4403. Springer, Berlin, pp 715–726, doi 10.1007/ 978-3-540-70928-2_54
Zitzler E, Künzli S (2004) Indicator-based selection in multiobjective search. In: Yao X, et al (eds) Conference on parallel problem solving from nature (PPSN VIII). LNCS, vol 3242. Springer, Berlin, pp 832–842
Zitzler E, Thiele L (1998) Multiobjective optimization using evolutionary algorithms – a comparative case study. In: Conference on parallel problem solving from nature (PPSN V), Amsterdam, pp 292–301
Zitzler E, Brockhoff D, Thiele L (2007) The hypervolume indicator revisited: on the design of Pareto-compliant indicators via weighted integration. In: Obayashi S, et al (eds) Conference on evolutionary multi-criterion optimization (EMO 2007). LNCS, vol 4403. Springer, Berlin, pp 862–876
Acknowledgements
The authors would like to thank Johannes Bader for the support with the illustrations. Dimo Brockhoff has been supported by the Swiss National Science Foundation (SNF) under grant 112079.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2010 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Brockhoff, D., Zitzler, E. (2010). Automated Aggregation and Omission of Objectives for Tackling Many-Objective Problems. In: Jones, D., Tamiz, M., Ries, J. (eds) New Developments in Multiple Objective and Goal Programming. Lecture Notes in Economics and Mathematical Systems, vol 638. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-10354-4_6
Download citation
DOI: https://doi.org/10.1007/978-3-642-10354-4_6
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-642-10353-7
Online ISBN: 978-3-642-10354-4
eBook Packages: Mathematics and StatisticsMathematics and Statistics (R0)