Comments (2)
@raphaelreinauer can you please provide a quick description of this issue? In particular, what preprocessing do you need and why, for example, you are not using the preprocessing tools of persistent diagrams available in giotto-tda?
from giotto-deep.
The issue with gtda.diagrams.preprocessing and sklearn.base.TransformerMixin is that they work pretty well when one can load the whole np.ndarray or torch.Tensor in memory, which is not my case here...I have lots of
persistent diagrams that I can't fit into my memory. Hence the sklearn paradigm of fit and transform to whole arrays or tensors does not apply here since I have to perform the operation in batches.
Hence I want to use the PyTorch or Huggingface paradigm of preprocessing the data when loading them in a data loader via PyTorch DataLoader. The main goal is to implement
a preprocessing module that I can apply to single persistence diagrams as well as to batches, allowing it to be used either in the getitem or collate_fn functions of pytorch.data.dataset.Dataset to create something like
(https://huggingface.co/docs/transformers/main_classes/feature_extractor). Furthermore, it should be able to be saved to memory and loaded from memory since the model I train depends on the preprocessing that is
applied to test training data. Hence it's crucial for inference to be able to load exactly the same preprocessing parameters in order to apply them to the input data as well.
I hope this answers your question. If not, I can try to be more specific.
from giotto-deep.
Related Issues (20)
- [Proposal] Add `pre-commit` HOT 2
- ImportError: cannot import name 'PILLOW_VERSION' from 'PIL' HOT 7
- Delete unused branches HOT 2
- build CI for docker build
- WIP: notes from JOSS review HOT 1
- Leftover code for print delay
- [BUG] Using torchvision's root parameter through DatasetBuilder causes error
- Accumulate gradients is not compatible with BatchNorm HOT 1
- [BUG] Properly define public api HOT 1
- [BUG] Usage of eval is bad practice: use factory design pattern HOT 2
- Generic preprocessing module HOT 2
- Add a tutorial on how to use Huggingface models with giotto-deep
- prepare to submit paper to JOSS HOT 4
- Create a persistence diagram type
- Template for Preprocessing Pipeline HOT 9
- Notebook on Orbit5k HOT 2
- modularise the Persformer
- Use python-RQ to distribute jobs
- Add GPUs to pods HOT 2
- Add detailed docstring to the OneHotEncodedPersistenceDiagram class
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 giotto-deep.