I have recently learnt about kernels in machine learning. And I have been introduced to many different processes e.g. Gaussian process, Wiener process. Now my question is why a set of functions has been named as a process? For example this is the definition of Gaussian process:

Let $\mu: X \to R$ be any function, $k: X \times X\to R$ be a Mercer kernel. A Gaussian process $p(f) = GP(f;μ,k)$ is a probability distribution over the function $f : X \to R$, such that every finite restriction to function values $f_X :=[f_{x_1},...,f_{x_N}]$ is a Gaussian distribution $p(f_X)=N(f_X;μ_X,k_{XX})$.

Which parts exactly associates to a process??


I believe it's called a process because it takes place over and can change over time. In the case of your examples, they are stochastic processes.

EDIT: As Benjamin says in a comment, the requirement isn't time, but that there is an indexing set, which could be time. It could also be spatial.

  • 1
    $\begingroup$ a process does not necessarily have to take place over/change over time, but there is some set that 'indexes' a collection of random variables. This set can be continuous (e.g. time) or discrete (e.g. a set of possible states like the 4 possible nucleotides in statistical genetics). $\endgroup$ – bdeonovic Feb 9 '14 at 20:17

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.