Question about Gaussian Process regression using GPML toolbox

Hello, I am using GPML toolbox for learning multiple motion models from my data and then using those models for making regression (prediction of locations). In the training phase, a set of GP motion models have is learned from the training data set. Then in the prediction phase, I need to choose the proper motion model first among all models and then use that for making prediction. So my question is for a test case, how can I decide between multiple models?

Thanks, Sepideh

Answer 1

[Cameron LaMack]

Hello!

While this isn't really a MATLAB question, I'm still happy to help. The answer is you (or some other machine learning classifier). I might be able to help more if I knew what your inputs and outputs were, but the short answer is that it sounds like you'll have to use expertise to select an appropriate model, then use that model to create predictions.

Hopefully this helps!