Clear Filters
Clear Filters

How to use observational weights in fitrgp?

21 views (last 30 days)
Andreas
Andreas on 2 Oct 2024
Commented: Umar on 9 Oct 2024 at 0:39
Is there a way to introduce observational weights in the Gaussian Process Regression (fitrgp) routine? This seems to be possible for other machine learning methods such as fitrensemble.
  7 Comments
Andreas
Andreas on 9 Oct 2024 at 0:12
Hi Umar,
First of all thank you very much for your very detailed response - much appreciated!
I read through your code example for the implementation steps for the custom loss function. What I'm not sure I understand is where the custom loss function is utilized during the model training. It appears to me that "weightedLoss" is only called after the training to evalute the model performance by means of a weighted MSE. This wouldn't affect the training process but only the post-training analytics. Is that right or am I missing anything?
Umar
Umar on 9 Oct 2024 at 0:39
Hi @Andreas,
You are correct in noting that the current implementation only evaluates performance post-training without affecting the training process itself. To effectively integrate observation weights into GPR modeling in MATLAB would require either workaround methods or switching to a platform that supports such functionality more natively.
Feel free to reach out if you have any further questions or need more clarification on this topic!

Sign in to comment.

Answers (0)

Tags

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!