Sparse random feature networks with in-degree d are equivalent to order-d additive models, and a distribution of in-degrees yields a mixture of additive kernels.
Avoiding pathologies in very deep networks
1 Pith paper cite this work. Polarity classification is still indexing.
abstract
Choosing appropriate architectures and regularization strategies for deep networks is crucial to good predictive performance. To shed light on this problem, we analyze the analogous problem of constructing useful priors on compositions of functions. Specifically, we study the deep Gaussian process, a type of infinitely-wide, deep neural network. We show that in standard architectures, the representational capacity of the network tends to capture fewer degrees of freedom as the number of layers increases, retaining only a single degree of freedom in the limit. We propose an alternate network architecture which does not suffer from this pathology. We also examine deep covariance functions, obtained by composing infinitely many feature transforms. Lastly, we characterize the class of models obtained by performing dropout on Gaussian processes.
citation-role summary
citation-polarity summary
fields
cs.NE 1years
2019 1verdicts
CONDITIONAL 1roles
background 1polarities
unclear 1representative citing papers
citing papers explorer
-
Additive function approximation in the brain
Sparse random feature networks with in-degree d are equivalent to order-d additive models, and a distribution of in-degrees yields a mixture of additive kernels.