Deep convolutional Gaussian processes
This addresses image classification for researchers and practitioners by offering a principled Bayesian alternative, though it appears incremental as it builds on existing Gaussian process approaches with convolutional enhancements.
The paper tackled image classification by proposing deep convolutional Gaussian processes, a Bayesian framework for hierarchical feature detection, resulting in over 10 percentage point accuracy improvement on CIFAR-10 compared to existing Gaussian process methods.
We propose deep convolutional Gaussian processes, a deep Gaussian process architecture with convolutional structure. The model is a principled Bayesian framework for detecting hierarchical combinations of local features for image classification. We demonstrate greatly improved image classification performance compared to current Gaussian process approaches on the MNIST and CIFAR-10 datasets. In particular, we improve CIFAR-10 accuracy by over 10 percentage points.