A framework for co-optimization algorithm performance and its application to worst-case optimization

Traditional black-box optimization searches a set of potential solutions for those optimizing the value of a function whose analytical or algebraic form is unknown or inexistent, but whose value can be queried for any input. Co-optimization is a generalization of this setting, in which fully evaluating a potential solution may require querying some function more than once, typically a very large number of times. When that's the case, co-optimization poses unique difficulties to designing and assessing algorithms. A generally-applicable approach is to judge co-optimization algorithm performance via an aggregate over all possible functions in the problem domain. We establish formal definitions of such aggregate performance and then investigate the following questions concerning algorithm design: 1) are some algorithms strictly better than others? i.e. is there "free lunch"? 2) do optimal algorithms exist? and 3) if so, are they practical? We formally define free lunch and aggregate optimality of co-optimization algorithms and derive generic conditions for their existence. We review and explain prior (no) free lunch results from the perspective of these conditions; we also show how this framework can be used to bridge several fields of research, by allowing formalization of their various problems and views on performance. We then apply and extend the generic results in a context involving a particular type of co-optimization called worst-case optimization. In this context we show that there exist algorithms that are aggregately-optimal for any budget (allowed number of function calls) and any starting point (set of previously uncovered function call outcomes), and also non-trivially strictly optimal for many budgets and starting points; moreover, we formalize the operation of such optimal algorithms and show that for certain domains, budgets and starting points this operation is equivalent to a simple procedure with tractable implementation-a first-of-its-kind result for co-optimization.

[1]  Richard S. Sutton,et al.  Introduction to Reinforcement Learning , 1998 .

[2]  Matthew J. Streeter,et al.  Two Broad Classes of Functions for Which a No Free Lunch Result Does Not Hold , 2003, GECCO.

[3]  David H. Wolpert,et al.  No free lunch theorems for optimization , 1997, IEEE Trans. Evol. Comput..

[4]  Mikkel T. Jensen.  Finding worst-case flexible schedules using coevolution , 2001 .

[5]  Edwin D. de Jong,et al.  Coevolutionary Principles , 2012, Handbook of Natural Computing.

[6]  Thomas Jansen,et al.  Design and Management of Complex Technical Processes and Systems by Means of Computational Intelligence Methods Perhaps Not a Free Lunch but at Least a Free Appetizer Perhaps Not a Free Lunch but at Least a Free Appetizer , 2022 .

[7]  Daniel R. Tauritz,et al.  A no-free-lunch framework for coevolution , 2008, GECCO '08.

[8]  Fredrik A. Dahl,et al.  Asymmetric Co-evolution for Imperfect-Information Zero-Sum Games , 2000, ECML.

[9]  Marc Toussaint,et al.  On Classes of Functions for which No Free Lunch Results Hold , 2001, Inf. Process. Lett..

[10]  Marc Toussaint,et al.  A No-Free-Lunch Theorem for Non-Uniform Distributions of Target Functions , 2004 .

[11]  Thomas Jansen,et al.  A method to derive fixed budget results from expected optimisation times , 2013, GECCO '13.

[12]  Thomas Jansen,et al.  Optimization with randomized search heuristics - the (A)NFL theorem, realistic scenarios, and difficult functions , 2002, Theor. Comput. Sci..

[13]  Thomas Jansen,et al.  Performance analysis of randomised search heuristics operating with a fixed budget , 2014, Theor. Comput. Sci..

[14]  V. F. Demʹi︠a︡nov,et al.  Introduction to minimax , 1976 .

[15]  Benjamin Doerr,et al.  Playing Mastermind with Constant-Size Memory , 2011, Theory of Computing Systems.

[16]  Thomas Jansen,et al.  Fixed budget computations: a different perspective on run time analysis , 2012, GECCO '12.

[17]  Elena Popovici,et al.  An analysis of two-population coevolutionary computation , 2006 .

[18]  Tom Everitt,et al.  Universal Induction and Optimisation: No Free Lunch , 2013 .

[19]  Travis C. Service.  Unbiased coevolutionary solution concepts , 2009, FOGA.

[20]  Kenneth A. De Jong,et al.  Monotonicity versus performance in co-optimization , 2009, FOGA '09.

[21]  F. Alajaji,et al.  c ○ Copyright by , 1998 .

[22]  Mikkel T. Jensen,et al.  Robust and Flexible Scheduling with Evolutionary Computation , 2001 .

[23]  V. F. Demʹi︠a︡nov,et al.  Introduction to minimax , 1976 .

[24]  Olivier Teytaud,et al.  Continuous Lunches Are Free Plus the Design of Optimal Optimization Algorithms , 2010, Algorithmica.

[25]  David W. Corne,et al.  No Free Lunch and Free Leftovers Theorems for Multiobjective Optimisation Problems , 2003, EMO.

[26]  L. D. Whitley,et al.  The No Free Lunch and problem description length , 2001 .

[27]  Jeffrey W. Herrmann,et al.  A genetic algorithm for minimax optimization problems , 1999, Proceedings of the 1999 Congress on Evolutionary Computation-CEC99 (Cat. No. 99TH8406).

[28]  Dirk Sudholt,et al.  A fixed budget analysis of randomized search heuristics for the traveling salesperson problem , 2014, GECCO.

[29]  David H. Wolpert,et al.  Coevolutionary free lunches , 2005, IEEE Transactions on Evolutionary Computation.

[30]  R. Paul Wiegand,et al.  Black-box search by elimination of fitness functions , 2009, FOGA '09.

[31]  H. Barbosa.  A coevolutionary genetic algorithm for constrained optimization , 1999, Proceedings of the 1999 Congress on Evolutionary Computation-CEC99 (Cat. No. 99TH8406).

[32]  Daniel R. Tauritz,et al.  Co-optimization algorithms , 2008, GECCO '08.

[33]  Gerald Tesauro,et al.  Temporal difference learning and TD-Gammon , 1995, CACM.

[34]  L. Darrell Whitley,et al.  A "No Free Lunch" Tutorial: Sharpened and Focused No Free Lunch , 2011, Theory of Randomized Search Heuristics.

[35]  Thomas Jansen,et al.  Black-Box Complexity for Bounding the Performance of Randomized Search Heuristics , 2014, Theory and Principled Methods for the Design of Metaheuristics.

[36]  Elena Popovici,et al.  On the practicality of optimal output mechanisms for co-optimization algorithms , 2011, FOGA '11.

[37]  A Gerodimos,et al.  Robust Discrete Optimization and its Applications , 1996, J. Oper. Res. Soc..

[38]  Daniel R. Tauritz,et al.  Free lunches in pareto coevolution , 2009, GECCO '09.

[39]  Helio J. C. Barbosa.  A Coevolutionary Genetic Algorithm for a Game Approach to Structural Optimization , 1997, ICGA.