A PAC-Bayesian Approach to Spectrally-Normalized Margin Bounds for Neural NetworksDownload PDF

Feb 15, 2018 (edited Feb 23, 2018)ICLR 2018 Conference Blind SubmissionReaders: Everyone
  • Abstract: We present a generalization bound for feedforward neural networks in terms of the product of the spectral norm of the layers and the Frobenius norm of the weights. The generalization bound is derived using a PAC-Bayes analysis.
  • Keywords: Neural Networks, Generalization, PAC-Bayes, Sharpness
4 Replies