Implement GridComparator #113
Conversation
|
|
||
| self.comparison_array[row, col, channel] = pair_result | ||
|
|
||
| return self.comparison_array |
There was a problem hiding this comment.
is there any kind of way we can give this a meaningful order?
There was a problem hiding this comment.
or, phrased differently, what is the use case that you have in mind?
There was a problem hiding this comment.
or, phrased differently, what is the use case that you have in mind?
It is essentially a multi-channel adjacency matrix. A set of adjacency matrices that store information on different types of relationships.
There was a problem hiding this comment.
is there any kind of way we can give this a meaningful order?
I am working on that. Presently, I have implemented a mapping between the factors of interest and some numerical indices. If I had to improve on this, I'd probably need something like xarray.
There was a problem hiding this comment.
What are your ideas for approaches for giving this a consistent shape? At least for most conventional ML approaches, you will need your inputs to always have the same shape (or you only limit yourself to producing sequences)
This PR should implement a GridComparator. This would allow the generation of similarity grids for sets of molecules.