First Steps Towards a Runtime Analysis When Starting With a Good Solution

Denis Antipov, Maxim Buzdalov, Benjamin Doerr

Allbwn ymchwil: Cyfraniad at gyfnodolynErthygladolygiad gan gymheiriaid

8 Wedi eu Llwytho i Lawr (Pure)

Crynodeb

The mathematical runtime analysis of evolutionary algorithms traditionally regards the time an algorithm needs to find a solution of a certain quality when initialized with a random population. In practical applications it may be possible to guess solutions that are better than random ones. We start a mathematical runtime analysis for such situations. We observe that different algorithms profit to a very different degree from a better initialization. We also show that the optimal parameterization of an algorithm can depend strongly on the quality of the initial solutions. To overcome this difficulty, self-adjusting and randomized heavy-tailed parameter choices can be profitable. Finally, we observe a larger gap between the performance of the best evolutionary algorithm we found and the corresponding black-box complexity. This could suggest that evolutionary algorithms better exploiting good initial solutions are still to be found. These first findings stem from analyzing the performance of the \((1+1)\) evolutionary algorithm and the static, self-adjusting, and heavy-tailed \((1+(\lambda,\lambda))\) genetic algorithms on the OneMax benchmark. We are optimistic that the question of how to profit from good initial solutions is interesting beyond these first examples.
Iaith wreiddiolSaesneg
CyfnodolynACM Transactions on Evolutionary Learning and Optimization
Dyddiad ar-lein cynnar01 Gorff 2024
Dynodwyr Gwrthrych Digidol (DOIs)
StatwsE-gyhoeddi cyn argraffu - 01 Gorff 2024

Ôl bys

Gweld gwybodaeth am bynciau ymchwil 'First Steps Towards a Runtime Analysis When Starting With a Good Solution'. Gyda’i gilydd, maen nhw’n ffurfio ôl bys unigryw.
  • First Steps Towards a Runtime Analysis When Starting with a Good Solution

    Antipov, D., Buzdalov, M. & Doerr, B., 02 Medi 2020, Parallel Problem Solving from Nature – PPSN XVI. Bäck, T., Preuss, M., Deutz, A., Emmerich, M., Wang, H., Doerr, C. & Trautmann, H. (gol.). Springer Nature, t. 560-573 14 t. (Lecture Notes in Computer Science; Cyfrol 12270).

    Allbwn ymchwil: Pennod mewn Llyfr/Adroddiad/Trafodion CynhadleddTrafodion Cynhadledd (Nid-Cyfnodolyn fathau)

    Mynediad agored
    Ffeil
    23 Dyfyniadau (Scopus)
    40 Wedi eu Llwytho i Lawr (Pure)

Dyfynnu hyn