A Gaussian Process perspective on Convolutional Neural Networks
This work addresses the theoretical understanding of CNNs for researchers, but it is incremental as it extends known Gaussian process results from fully-connected networks to convolutional ones.
The paper tackles the problem of understanding convolutional neural networks (CNNs) by casting them into a Gaussian process perspective, aiming to gain insights into their performance and implicit assumptions, with a focus on when their output approaches a multivariate normal distribution.
In this paper we cast the well-known convolutional neural network in a Gaussian process perspective. In this way we hope to gain additional insights into the performance of convolutional networks, in particular understand under what circumstances they tend to perform well and what assumptions are implicitly made in the network. While for fully-connected networks the properties of convergence to Gaussian processes have been studied extensively, little is known about situations in which the output from a convolutional network approaches a multivariate normal distribution.