Comments (5)
I'm also confused about this. if args.random_rotation, the transforms are passed to the training dataset. however, inside the training loop, the same transforms are done again, but looping each sample inside the batch.
It seems inefficient, and also quite confusing, because I also realised the rotation operations inside iterate_surface_precompute
are using P1["rand_rot"].T
while the rotation operations in the main training loop (data_iteration.iterate
) are using P1["rand_rot"]
. So are we actually reversing the rotation ...? (then why do we rotate in the first place?)
from dmasif.
I guess that these augmentations are used only for surface precomputation. So, they are undone in the main training loop. It is rather strange, but i get it.
from dmasif.
I guess that these augmentations are used only for surface precomputation. So, they are undone in the main training loop. It is rather strange, but i get it.
Could you please clarify why this makes sense? Why do we need to undo the augmentation? Don't we want the model to not care about the rotation & center positions?
from dmasif.
I guess that these augmentations are used only for surface precomputation. So, they are undone in the main training loop. It is rather strange, but i get it.
Could you please clarify why this makes sense? Why do we need to undo the augmentation? Don't we want the model to not care about the rotation & center positions?
I think, rotations and translations do not matter, when you use masif. But they also use dgcnn for comparison, and this model does not have rotational invariance, so augmentations in the main training loop are necessary. That is why it is strange that they do not use them.
from dmasif.
@FreyrS Hi Frey, could you provide some clarification to this part? It would be very helpful for us
from dmasif.
Related Issues (20)
- Choice of the atom RADII HOT 1
- How to predict the interface characters of single protein? HOT 1
- dMaSIF for interaction prediction, how to find complementary regions? HOT 2
- No save model checkpoints for site prediction HOT 2
- Input seems to require a y value HOT 2
- Running you example does not provide same ROC-AUC as claimed in the paper HOT 14
- Using main_training.py
- Problems regarding loss function and auroc
- Does anyone have a docker image that can run this project? Can you share it with me? HOT 4
- How to replace with my own dataset
- why the interface is computed like this ? HOT 4
- About helper.py
- Pre-trained model in Hugging face
- [Question] Are ligands or PTMs considered
- How to mathematically define the binding surface? HOT 1
- Why "iface_preds " contain NAN when training dmasif_ HOT 7
- Return value in diagonal_ranges function
- embedding example HOT 1
- faces of the protein surface
- RuntimeError: operation does not have an identity.
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from dmasif.