Frequency Fitness Assignment

被引:14
|
作者
Weise, Thomas [1 ]
Wan, Mingxu [1 ]
Wang, Pu [1 ]
Tang, Ke [1 ]
Devert, Alexandre [1 ]
Yao, Xin [1 ,2 ]
机构
[1] Univ Sci & Technol China, Sch Comp Sci & Technol, USTC Birmingham Joint Res Inst Intelligent Comput, Hefei 230027, Anhui, Peoples R China
[2] Univ Birmingham, Sch Comp Sci, Ctr Excellence Res Computat Intelligence & Applic, Birmingham B15 2TT, W Midlands, England
基金
中国国家自然科学基金;
关键词
Combinatorial optimization; diversity; fitness assignment; frequency; genetic programming (GP); numerical optimization; ALGORITHM;
D O I
10.1109/TEVC.2013.2251885
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Metaheuristic optimization procedures such as evolutionary algorithms are usually driven by an objective function that rates the quality of a candidate solution. However, it is not clear in practice whether an objective function adequately rewards intermediate solutions on the path to the global optimum and it may exhibit deceptiveness, epistasis, neutrality, ruggedness, and a lack of causality. In this paper, we introduce the frequency fitness H, subject to minimization, which rates how often solutions with the same objective value have been discovered so far. The ideas behind this method are that good solutions are difficult to find and that if an algorithm gets stuck at a local optimum, the frequency of the objective values of the surrounding solutions will increase over time, which will eventually allow it to leave that region again. We substitute a frequency fitness assignment process (FFA) for the objective function into several different optimization algorithms. We conduct a comprehensive set of experiments: the synthesis of algorithms with genetic programming (GP), the solution of MAX-3SAT problems with genetic algorithms, classification with Memetic Genetic Programming, and numerical optimization with a (1+1) Evolution Strategy, to verify the utility of FFA. Given that they have no access to the original objective function at all, it is surprising that for some problems (e. g., the algorithm synthesis task) the FFA-based algorithm variants perform significantly better. However, this cannot be guaranteed for all tested problems. Thus, we also analyze scenarios where algorithms using FFA do not perform better or perform even worse than with the original objective functions.
引用
收藏
页码:226 / 243
页数:18
相关论文
共 50 条
  • [1] Entropy, Search Trajectories, and Explainability for Frequency Fitness Assignment
    Thomson, Sarah L.
    Ochoa, Gabriela
    van den Berg, Daan
    Liang, Tianyu
    Weise, Thomas
    PARALLEL PROBLEM SOLVING FROM NATURE-PPSN XVIII, PPSN 2024, PT I, 2024, 15148 : 377 - 392
  • [2] Solving the Traveling Salesperson Problem using Frequency Fitness Assignment
    Liang, Tianyu
    Wu, Zhize
    Laessig, Jörg
    van den Berg, Daan
    Weise, Thomas
    2022 IEEE SYMPOSIUM SERIES ON COMPUTATIONAL INTELLIGENCE (SSCI), 2022, : 360 - 367
  • [3] Addressing the traveling salesperson problem with frequency fitness assignment and hybrid algorithms
    Liang T.
    Wu Z.
    Lässig J.
    Berg D.V.
    Thomson S.L.
    Weise T.
    Soft Computing, 2024, 28 (17-18) : 9495 - 9508
  • [4] Frequency Fitness Assignment: Optimization Without Bias for Good Solutions Can Be Efficient
    Weise, Thomas
    Wu, Zhize
    Li, Xinlu
    Chen, Yan
    Laessig, Joerg
    IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION, 2023, 27 (04) : 980 - 992
  • [5] Hybrid fitness assignment strategy in IGA - A method to compose fitness
    Sugimoto, F
    Yoneyama, M
    PROCEEDINGS OF THE 2002 IEEE WORKSHOP ON MULTIMEDIA SIGNAL PROCESSING, 2002, : 284 - 287
  • [6] Frequency Fitness Assignment: Making Optimization Algorithms Invariant Under Bijective Transformations of the Objective Function Value
    Weise, Thomas
    Wu, Zhize
    Li, Xinlu
    Chen, Yan
    IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION, 2021, 25 (02) : 307 - 319
  • [7] On the assignment of fitness to parents and offspring: whose fitness is it and when does it matter?
    Wolf, JB
    Wade, MJ
    JOURNAL OF EVOLUTIONARY BIOLOGY, 2001, 14 (02) : 347 - 356
  • [8] AN APPROACH TO FREQUENCY ASSIGNMENT
    FREEMAN, ER
    IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY, 1966, EMC8 (02) : 90 - &
  • [9] OPTIMIZATION OF FREQUENCY ASSIGNMENT
    MIZUIKE, T
    ITO, Y
    IEEE TRANSACTIONS ON COMMUNICATIONS, 1989, 37 (10) : 1031 - 1041
  • [10] On the assignment of fitness values in statistical analyses of selection
    Brodie, ED
    Janzen, FJ
    EVOLUTION, 1996, 50 (01) : 437 - 442