Choosing kernels in GPs

In many applications, we find that the prior mean is always set to zero as it is simple, convenient and works well for many applications. However, choosing the appropriate kernels for the task is not always straightforward. As mentioned in the previous section, kernels effectively try to map the similarity between two input data points in the input space to the output (function) space. The only requirement for the kernel function () is that it should map any two input values  and  to a scalar such that Kernel Matrix ( ...

Get TensorFlow Machine Learning Projects now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.