Corpus ID: 220128109

Training Convolutional ReLU Neural Networks in Polynomial Time: Exact Convex Optimization Formulations

@article{Ergen2020TrainingCR,
  title={Training Convolutional ReLU Neural Networks in Polynomial Time: Exact Convex Optimization Formulations},
  author={Tolga Ergen and Mert Pilanci},
  journal={ArXiv},
  year={2020},
  volume={abs/2006.14798}
}
  • Tolga Ergen, Mert Pilanci
  • Published 2020
  • Computer Science, Mathematics
  • ArXiv
  • We study training of Convolutional Neural Networks (CNNs) with ReLU activations and introduce exact convex optimization formulations with a polynomial complexity with respect to the number of data samples, the number of neurons and data dimension. Particularly, we develop a convex analytic framework utilizing semi-infinite duality to obtain equivalent convex optimization problems for several CNN architectures. We first prove that two-layer CNNs can be globally optimized via an l2 norm… CONTINUE READING

    References

    Publications referenced by this paper.
    SHOWING 1-10 OF 33 REFERENCES

    Breaking the Curse of Dimensionality with Convex Neural Networks

    VIEW 2 EXCERPTS

    CVX: Matlab software for disciplined convex programming, version 2.1

    • Michael Grant, Stephen Boyd
    • 2014
    VIEW 2 EXCERPTS

    CVXPY: A Python-Embedded Modeling Language for Convex Optimization

    VIEW 2 EXCERPTS

    Constructing arrangements of lines and hyperplanes with applications