The article presents Neural Operator Variational Inference (NOVI) for Deep Gaussian Processes (DGP). NOVI uses a neural generator to obtain a sampler and minimizes the Regularized Stein Discrepancy between the generated distribution and true posterior. This approach addresses challenges in exact inference and overcomes computational expenses of stochastic approximation. The proposed method demonstrated faster convergence rates and achieved a classification accuracy of 93.56 on the CIFAR10 dataset, outperforming state-of-the-art Gaussian process methods. NOVI has potential implications for enhancing the performance of deep Bayesian nonparametric models in various applications.
Publication date: 25 Sep 2023
Project Page: Not provided
Paper: https://arxiv.org/pdf/2309.12658
