A Study of Fitness Landscapes for Neuroevolution

@article{Rodrigues2020ASO,
  title={A Study of Fitness Landscapes for Neuroevolution},
  author={Nuno M. M. Rodrigues and S. Silva and L. Vanneschi},
  journal={2020 IEEE Congress on Evolutionary Computation (CEC)},
  year={2020},
  pages={1-8}
}
  • Nuno M. M. Rodrigues, S. Silva, L. Vanneschi
  • Published 2020
  • Computer Science
  • 2020 IEEE Congress on Evolutionary Computation (CEC)
  • Fitness landscapes are a useful concept to study the dynamics of meta-heuristics. In the last two decades, they have been applied with success to estimate the optimization power of several types of evolutionary algorithms, including genetic algorithms and genetic programming. However, so far they have never been used to study the performance of machine learning algorithms on unseen data, and they have never been applied to neuroevolution. This paper aims at filling both these gaps, applying for… CONTINUE READING

    References

    SHOWING 1-10 OF 31 REFERENCES
    Evolving Neural Networks through Augmenting Topologies
    • 2,581
    • PDF
    Correlated and uncorrelated fitness landscapes and how to tell the difference
    • E. Weinberger
    • Mathematics, Computer Science
    • Biological Cybernetics
    • 2004
    • 471
    Theory and practice for efficient genetic programming
    • 91
    A Hypercube-Based Encoding for Evolving Large-Scale Neural Networks
    • 610
    Quantifying ruggedness of continuous landscapes using entropy
    • 69
    Fitness Distance Correlation as a Measure of Problem Difficulty for Genetic Algorithms
    • 748
    • PDF
    Fitness landscapes and graphs: multimodularity, ruggedness and neutrality
    • S. Vérel
    • Mathematics, Computer Science
    • GECCO '12
    • 2012
    • 5
    • PDF
    A new evolutionary system for evolving artificial neural networks
    • X. Yao, Yong Liu
    • Computer Science, Medicine
    • IEEE Trans. Neural Networks
    • 1997
    • 900
    • PDF
    Analysis of error landscapes in multi-layered neural networks for classification
    • 8