Deep learning with differential Gaussian process flows
This work addresses the need for more effective and generalizable deep learning architectures, offering a novel paradigm that could benefit researchers and practitioners in machine learning.
The authors tackled the problem of improving deep learning models by proposing differential Gaussian process flows, which transform inputs through stochastic differential equations before classification or regression, achieving state-of-the-art results that outperform deep Gaussian processes and neural networks.
We propose a novel deep learning paradigm of differential flows that learn a stochastic differential equation transformations of inputs prior to a standard classification or regression function. The key property of differential Gaussian processes is the warping of inputs through infinitely deep, but infinitesimal, differential fields, that generalise discrete layers into a dynamical system. We demonstrate state-of-the-art results that exceed the performance of deep Gaussian processes and neural networks