About model generalization

6 views
Skip to first unread message

Bereket Workie Anteneh

unread,
Aug 25, 2021, 11:16:48 AMAug 25
to scalismo
Dear Scalismo Community
I want to do generalization ability of a model using leave one out cross validation test. 
I tried to see the way of implementation from former messages in the forum but I am not fully clear. I think conceptually, generalization ability is  implemented by constructing a SSM by excluding one training mesh and then reconstructing or fitting the excluded shape objects with m principal components. The process is repeated for all training examples. 
On this issue I have two questions.
The first is: to create SSM the meshes should be in correspondence (equal number of points), i.e. the meshes are in correspondence means they are fitted to each other. So, as much as they are fitted to each other why we did reconstruction of the excluded shape with the built model?
Second is: how I can implement this concept into practical (code)? 

With regards,
Reply all
Reply to author
Forward
0 new messages