Entropy, Search Trajectories, and Explainability for Frequency Fitness Assignment

被引:0
|
作者
Thomson, Sarah L. [1 ]
Ochoa, Gabriela [2 ]
van den Berg, Daan [3 ]
Liang, Tianyu [4 ]
Weise, Thomas [4 ]
机构
[1] Edinburgh Napier Univ, Edinburgh, Midlothian, Scotland
[2] Univ Stirling, Stirling, Scotland
[3] Vrije Univ, Amsterdam, Netherlands
[4] Hefei Univ, Hefei, Peoples R China
来源
PARALLEL PROBLEM SOLVING FROM NATURE-PPSN XVIII, PPSN 2024, PT I | 2024年 / 15148卷
关键词
frequency fitness assignment; quadratic assignment problem; fitness landscape; APPROXIMATE ENTROPY;
D O I
10.1007/978-3-031-70055-2_23
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Local optima are a menace that can trap optimisation processes. Frequency fitness assignment (FFA) is an concept aiming to overcome this problem. It steers the search towards solutions with rare fitness instead of high-quality fitness. FFA-based algorithms have shown promise in the literature, but their behaviour is not well understood. We take a first step in this direction by seeking to explain FFA behaviour and performance for the first time. In particular, we attempt to understand the difference in how FFA-based algorithms navigate the space when compared with a standard objective-guided algorithm which incorporates diversification: simulated annealing (SA). As a testbed for these investigations, a set of quadratic assignment problem (QAP) benchmark instances designed to be difficult for metaheuristics is used. A statistical analysis of trajectory behaviours for FFA-based algorithms is conducted. Additionally, we consider and compare the fitness distributions encountered by each algorithm, as well their respective proficiency on the problems. The findings help to explain FFA performance behaviours, and show that FFA explores more widely and consistently than SA. It is hoped that the explanatory approach adopted in this study serves as an example and inspires further similar investigations into how-and why-FFA-assisted optimisation works.
引用
收藏
页码:377 / 392
页数:16
相关论文
共 50 条
  • [41] Predictability of Evolutionary Trajectories in Fitness Landscapes
    Lobkovsky, Alexander E.
    Wolf, Yuri I.
    Koonin, Eugene V.
    PLOS COMPUTATIONAL BIOLOGY, 2011, 7 (12)
  • [42] Evolutionary trajectories in rugged fitness landscapes
    Jain, K
    Krug, J
    JOURNAL OF STATISTICAL MECHANICS-THEORY AND EXPERIMENT, 2005,
  • [43] Swarm Intelligence, Scatter Search and Genetic Algorithm to Tackle a Realistic Frequency Assignment Problem
    Chaves-Gonzalez, Jose M.
    Vega-Rodriguez, Miguel A.
    Gomez-Pulido, Juan A.
    Sanchez-Perez, Juan M.
    DISTRIBUTED COMPUTING AND ARTIFICIAL INTELLIGENCE, 2010, 79 : 441 - 448
  • [44] An improved tabu search algorithm for the fixed-spectrum frequency-assignment problem
    Montemanni, R
    Moon, JNJ
    Smith, DH
    IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, 2003, 52 (04) : 891 - 901
  • [45] An heuristic search technique for fixed frequency assignment in non-homogeneous demand systems
    Jose-Revuelta, L. M. San
    SIGNAL PROCESSING, 2008, 88 (06) : 1461 - 1476
  • [46] Stochastic local search algorithms for graph set T-colouring and frequency assignment
    Chiarandini, Marco
    Stuetzle, Thomas
    CONSTRAINTS, 2007, 12 (03) : 371 - 403
  • [47] Stochastic Local Search Algorithms for Graph Set T-colouring and Frequency Assignment
    Marco Chiarandini
    Thomas Stützle
    Constraints, 2007, 12 : 371 - 403
  • [48] Generic Hardness Estimation using Fitness and Parameter Landscapes applied to Robust Taboo Search and the Quadratic Assignment Problem
    Pitzer, Erik
    Beham, Andreas
    Affenzeller, Michael
    PROCEEDINGS OF THE FOURTEENTH INTERNATIONAL CONFERENCE ON GENETIC AND EVOLUTIONARY COMPUTATION COMPANION (GECCO'12), 2012, : 393 - 400
  • [49] 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
  • [50] 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