Thank you for sharing this pointnet repository! It has been really useful to me!
As part of my masters research, I wanted to use it for binary classification of point clouds for neutrino data (neutrino point clouds versus noise point clouds). I was wondering if you had any comments on the required modifications to the existing architecture?
I have a basic understanding of deep learning and think I should be changing the activation function to sigmoid and using a binary cross entropy loss function as well. Are there any other factors I should be aware of? Any advice would be great! I have two classes as I mentioned previously with a large number of samples. I was able to run the code but got inaccurate results since I did not modify the architecture for binary classification!
Thank you for sharing this pointnet repository! It has been really useful to me!
As part of my masters research, I wanted to use it for binary classification of point clouds for neutrino data (neutrino point clouds versus noise point clouds). I was wondering if you had any comments on the required modifications to the existing architecture?
I have a basic understanding of deep learning and think I should be changing the activation function to sigmoid and using a binary cross entropy loss function as well. Are there any other factors I should be aware of? Any advice would be great! I have two classes as I mentioned previously with a large number of samples. I was able to run the code but got inaccurate results since I did not modify the architecture for binary classification!
Thank you!