No Free Lunch and Free Leftovers Theorems for Multiobjective Optimisation Problems
نویسندگان
چکیده
The classic NFL theorems are invariably cast in terms of single objective optimization problems. We confirm that the classic NFL theorem holds for general multiobjective fitness spaces, and show how this follows from a ‘single-objective’ NFL theorem. We also show that, given any particular Pareto Front, an NFL theorem holds for the set of all multiobjective problems which have that Pareto Front. It follows that, given any ‘shape’ or class of Pareto fronts, an NFL theorem holds for the set of all multiobjective problems in that class. These findings have salience in test function design. Such NFL results are cast in the typical context of absolute performance, assuming a performance metric which returns a value based on the result produced by a single algorithm. But, in multiobjective search we commonly use comparative metrics, which return performance measures based non-trivially on the results from two (or more) algorithms. Closely related to but extending the observations in the seminal NFL work concerning minimax distinctions between algorithms, we provide a ‘Free Leftovers’ theorem for comparative performance of algorithms over permutation functions; in words: over the space of permutation problems, every algorithm has some companion algorithm(s) which it outperforms, according to a certain well-behaved metric, when comparative performance is summed over all problems in the space.
منابع مشابه
Arbitrary function optimisation with metaheuristics - No free lunch and real-world problems
No free lunch theorems for optimisation suggest that empirical studies on benchmarking problems are pointless, or even cast negative doubts, when algorithms are being applied to other problems not clearly related to the previous ones. Roughly speaking, reported empirical results are not just the result of algorithms’ performances, but the benchmark used therein as well; and consequently, recomm...
متن کاملSome Multiobjective Optimizers are Better than Others
The No-Free-Lunch (NFL) theorems hold for general multiobjective fitness spaces, in the sense that, over a space of problems which is closed under permutation, any two algorithms will produce the same set of multiobjective samples. However, there are salient ways in which NFL does not generally hold in multiobjective optimization. Previously we have shown that a ‘free lunch’ can arise when comp...
متن کاملThe No Free Lunch Theorems for Optimisation: An Overview
Many algorithms have been devised for tackling combinatorial optimisation problems (COPs). Traditional Operations Research (OR) techniques such as Branch and Bound and Cutting Planes Algorithms can, given enough time, guarantee an optimal solution as they explicitly exploit features of the optimisation function they are solving. Specialised heuristics exist for most COPs that also exploit featu...
متن کاملOn the Benchmarking of Multiobjective Optimization Algorithm
The ”No Free Lunch” (NFL) theorems state that in average each algorithm has the same performance, when no a priori knowledge of single-objective cost function f is assumed. This paper extends the NFL theorems to the case of multi-objective optimization. Further it is shown that even in cases of a priori knowledge, when the performance measure is related to the set of extrema points sampled so f...
متن کاملOptimization, block designs and No Free Lunch theorems
We study the precise conditions under which all optimisation strategies for a given family of finite functions yield the same expected maximisation performance, when averaged over a uniform distribution of the functions. In the case of bounded-length searches in a family of Boolean functions, we provide tight connections between such “No Free Lunch” conditions and the structure of t-designs and...
متن کامل