Analysis of error landscapes in multi-layered neural networks for classification

@article{Rakitianskaia2016AnalysisOE,
  title={Analysis of error landscapes in multi-layered neural networks for classification},
  author={A. Rakitianskaia and E. Bekker and K. M. Malan and A. Engelbrecht},
  journal={2016 IEEE Congress on Evolutionary Computation (CEC)},
  year={2016},
  pages={5270-5277}
}
  • A. Rakitianskaia, E. Bekker, +1 author A. Engelbrecht
  • Published 2016
  • Computer Science
  • 2016 IEEE Congress on Evolutionary Computation (CEC)
  • Artificial neural networks are inherently high-dimensional, which limits our ability to visualise and understand their inner workings. Neural network architecture and training algorithm parameters are usually optimised on an ad hoc basis, with very limited insight into the nature of the objective function landscape. This study proposes using fitness landscape analysis to quantify topological properties of neural network error landscapes. Five techniques from the fitness landscape analysis field… CONTINUE READING
    Progressive gradient walk for neural network fitness landscape analysis
    • 4
    Fitness Landscape Analysis of Weight-Elimination Neural Networks
    • 3
    Search space boundaries in neural network error landscape analysis
    • 5
    • Open Access
    Visualising Basins of Attraction for the Cross-Entropy and the Squared Error Neural Network Loss Functions
    • 7
    • Open Access
    A Study of Generalization and Fitness Landscapes for Neuroevolution
    A Study of Fitness Landscapes for Neuroevolution

    References

    Publications referenced by this paper.
    SHOWING 1-10 OF 33 REFERENCES
    Multi-layer Perceptron Error Surfaces: Visualization, Structure and Modelling
    • 61
    • Highly Influential
    • Open Access
    Quantifying ruggedness of continuous landscapes using entropy
    • 66
    Fitness landscape analysis and memetic algorithms for the quadratic assignment problem
    • 413
    Designing a neural network for forecasting financial and economic time series
    • 935
    • Open Access
    Evolutionary Algorithms, Fitness Landscapes and Search
    • 457
    • Open Access
    Fitness Distance Correlation of Neural Network Error Surfaces: A Scalable, Continuous Optimization Problem
    • 10
    • Open Access
    Fitness Distance Correlation as a Measure of Problem Difficulty for Genetic Algorithms
    • 731
    • Open Access
    A survey of techniques for characterising fitness landscapes and some possible ways forward
    • 133
    • Open Access