View on GitHub

BMOBench

Black-Box Multi-Objective Optimization Benchmarking Platform

Download this project as a .zip file Download this project as a tar.gz file

In this platform, we address multi-objective optimization(minimization) problem:

Several algorithms/techniques have been proposed and studied to solve such problems. With this context, these algorithms are assessed in one of two ways, viz. theoretical, and empirical analysis. In theoretical analysis, a principled methodology is carried out to derive an analytical bound of the (run-time) solution quality. After t evaluations/steps, the quality of the returned solution is evaluated by a loss/regret measure.

Alternatively, empirical analysis employs experimental simulations of the algorithm on complex problems, gaining an insight on the algorithm’s practicality/applicability on real-world problems. With this regard, most of the time, methods proposed to solve MOPs are benchmarked on a different set of problems under arbitrary budgets of function evaluation. We are interested in empirically assessing published/novel multi-objective optimization algorithms in a unified (constantly updated) framework.

We invite the multi-objective community to test their published/novel algorithms in solving 100 MOPs reported in the literature where the feasible decision space has simple bound constraints, i.e., problems for which X=[l,u] and l<u. The benchmark validates the efficacy of the algorithms by computing several quality indicators which are reported in terms of data profiles.

IMPORTANT DATES

Brief Description

Paper Submission:

Getting Started with BMOBench

References

  1. A. Al-Dujaili and S. Suresh, “BMOBench: Black-box multiobjective optimization benchmarking platform,ArXiv e-prints, vol. arXiv:1605.07009, 2016.

  2. Custódio, Ana Luísa, et al. "Direct multisearch for multiobjective optimization." SIAM Journal on Optimization 21.3 (2011): 1109-1140.

  3. Brockhoff, Dimo, Thanh-Do Tran, and Nikolaus Hansen. "Benchmarking numerical multiobjective optimizers revisited." Genetic and Evolutionary Computation Conference (GECCO 2015). 2015.