AABI 2021, 3rd Symposium on Advances in Approximate Bayesian Inference, January-February 2021 (Virtual Event)
The Bayesian treatment of neural networks dictates that a prior distribution is considered
over the weight and bias parameters of the network. The non-linear nature of the model
implies that any distribution of the parameters has an unpredictable effect on the distribution of the function output. Gaussian processes offer a rigorous framework to define prior distributions over the space of functions. Our proposal is to impose such functional priors on well-established architectures of neural networks by means of minimising the Wasserstein distance between samples of stochastic processes. Early experimental results demonstrate the potential of functional priors for Bayesian neural networks.
© EURECOM. Personal use of this material is permitted. The definitive version of this paper was published in AABI 2021, 3rd Symposium on Advances in Approximate Bayesian Inference, January-February 2021 (Virtual Event) and is available at :