On the practicality of optimal output mechanisms for co-optimization algorithms

@inproceedings{Popovici2011OnTP,
  title={On the practicality of optimal output mechanisms for co-optimization algorithms},
  author={Elena Popovici and Ezra Winston and Anthony Bucci},
  booktitle={FOGA '11},
  year={2011}
}
Co-optimization problems involve one or more search spaces and a means of assessing interactions between entities in these spaces. Assessing a potential solution requires aggregating in some way the outcomes of a very large or infinite number of such interactions. This layer of complexity presents difficulties for algorithm design that are not encountered in ordinary optimization. For example, what a co-optimization algorithm should output is not at all obvious. Theoretical research has shown… 

Figures and Tables from this paper

Bridging Supervised Learning and Test-Based Co-optimization
TLDR
This paper takes a close look at the important commonalities and subtle differences between the well-established field of supervised learning and the much younger one of cooptimization, providing a two-way dictionary for the respective terminologies used to describe these concepts.
Visualizing adversaries : transparent pooling approaches for decision support in cybersecurity
TLDR
ESTABLO was beneficial to most problems as different variants dominated in different scenarios and POOLING was able to consistently produce individuals that performed well against adversaries and in the context of all of their peers.
A framework for co-optimization algorithm performance and its application to worst-case optimization
Co-Optimization Free Lunches: Tractability of Optimal Black-Box Algorithms for Maximizing Expected Utility
  • E. Popovici
  • Computer Science
    Evolutionary Computation
  • 2018
TLDR
The design of generally well-performing black-box algorithms for expected-utility maximization problems is interested, that is, algorithms which have access to the utility function only via input–output queries.

References

SHOWING 1-10 OF 23 REFERENCES
Monotonicity versus performance in co-optimization
TLDR
This analysis points out that even the weakest monotonicity may simply not be within reach for some solution concepts that are nonetheless of practical interest, and hypothesis put forward, in light of recent ``free lunch'' results, is that the authors may be faced with a difficult choice betweenmonotonicity and performance.
No free lunch theorems for optimization
A framework is developed to explore the connection between effective optimization algorithms and the problems they are solving. A number of "no free lunch" (NFL) theorems are presented which
Coevolutionary free lunches
TLDR
This paper presents a general framework covering most optimization scenarios and shows that in self-play there are free lunches: in coevolution some algorithms have better performance than other algorithms, averaged across all possible problems.
A genetic algorithm for minimax optimization problems
  • J. Herrmann
  • Computer Science
    Proceedings of the 1999 Congress on Evolutionary Computation-CEC99 (Cat. No. 99TH8406)
  • 1999
TLDR
Experimental results show that the two-space genetic algorithm can find robust solutions for minimax optimization problems, and this paper uses the algorithm to solve a parallel machine scheduling problem with uncertain processing times.
Robust and Flexible Scheduling with Evolutionary Computation
TLDR
This thesis presents two fundamentally different approaches for scheduling job shops facing machine breakdowns and a state of the art method for stochastic scheduling based on an idea of minimising the cost of a neighbourhood of schedules.
New Approaches to Coevolutionary Worst-Case Optimization
TLDR
This paper proposes a number of new variants of coevolutionary algorithms to co-evolve the worst case test cases along with the solution candidates, and shows that these techniques outperform previously proposed coEVolutionary worst-case optimizers on some simple test problems.
Unbiased coevolutionary solution concepts
The primary concern of this paper is the question: Is the goal of an optimal well behaved coevolutionary algorithm attainable? We approach this question from the point of view of the No Free Lunch
An analysis of two-population coevolutionary computation
TLDR
This dissertation is the first study that "glues" all four pieces together and provides a more holistic perspective of the field of CoEC by identifying a problem property and introducing tools for analyzing this property that are applicable across subareas.
Finding worst-case flexible schedules using coevolution
TLDR
In this paper a minimax formulation is used to develop a coevolutionary algorithm for finding worst case flexible schedules and this approach is compared to a standard scheduling approach and concluded to produce more flexible schedules.
Coevolutionary search among adversaries
TLDR
New methods are described that overcome these Aaws and make coevolution more efficient, able to solve several game learning test problems that cannot be efficiently solved without them.
...
...