Programatically Understand Gaussian Processes: Part 1(nipunbatra.github.io)
nipunbatra.github.io
Programatically Understand Gaussian Processes: Part 1
https://nipunbatra.github.io/blog/2019/GP-1.html
2 comments
I agree that distill.pub is great and so is the article you've mentioned. I did seek inspiration from the linked article amongst many others.
I'm kinda confused by the graphs like this: https://nipunbatra.github.io/blog/2019/20d-conditional.gif
I would expect the graphs to curve (somewhat) smoothly between the 3rd and 4th known points, but instead the 4th point shows up as this weird discontinuity—the graphs are far above it, and then spike down to the 4th point only very close to that point. Why is that? The result doesn't seem (to my untrained eyes) to be very satisfactory.
I would expect the graphs to curve (somewhat) smoothly between the 3rd and 4th known points, but instead the 4th point shows up as this weird discontinuity—the graphs are far above it, and then spike down to the 4th point only very close to that point. Why is that? The result doesn't seem (to my untrained eyes) to be very satisfactory.
Good questions. A few quick thoughts. These may not be enough though!
First, the mean function specified in the prior is zero for all the points. Thus, points would have a high density near zero.
I have a strong hunch that choosing a "smoother" kernel would not lead to this discontinuity.
First, the mean function specified in the prior is zero for all the points. Thus, points would have a high density near zero.
I have a strong hunch that choosing a "smoother" kernel would not lead to this discontinuity.
https://distill.pub/2019/visual-exploration-gaussian-process...
Everything from distill.pub is great.