Publication | Closed Access
A PAC-Bayesian Approach to Spectrally-Normalized Margin Bounds for Neural Networks
253
Citations
0
References
2017
Year
Mathematical ProgrammingGeneralization BoundModel OptimizationNeural Scaling LawEngineeringMachine LearningComputational Learning TheoryUncertainty QuantificationPattern RecognitionAi FoundationStatistical InferenceComputer ScienceSpectral NormStatistical Learning TheoryDeep LearningSupervised LearningFrobenius NormPac-bayesian Approach
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.