2c463dfdde588f3bfc60d53118c10d6b-AuthorFeedback.pdf

Neural Information Processing Systems 

About the performance gap of SGPs, the optimal variational posterior is indeed a Gaussian for single-layer SGP24 regression(Titsias,2009). Based on your comments for the MNIST experiment, we increase the number of inducing points to800.

Similar Docs  Excel Report  more

TitleSimilaritySource
None found