Stuck on an issue?

Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. It collects links to all the places you might be looking at while hunting down a tough bug.

And, if you’re still stuck at the end, we’re happy to hop on a call to see how we can help out.

Implement Nystroem.inverse_transform

See original GitHub issue

Our KernelPCA(fit_inverse_transform=True) implements the method of:

Learning to Find Pre-Images Neural Information Processing Systems 16 (NIPS 2003) Jason Weston, Bernhard Schölkopf, Gökhan Bakir

https://papers.nips.cc/paper/2003/hash/ac1ad983e08ad3304a97e147f522747e-Abstract.html

We might want to adapt this to make make_pipeline(Nystroem(), PCA()) a feature complete, scalable alternative to KernelPCA for denoising applications and novelty detection for instance.

Issue Analytics

State:
Created 2 years ago
Comments:6 (6 by maintainers)

Top GitHub Comments

1reaction

glemaitrecommented, Apr 20, 2021

@ogrisel or I did not start to look at it yet. Feel free to open a PR and we will be happy to give feedback. It would probably be interesting to have a couple of benchmark to check the computational and statistical performance of such approach 😃 Looking forward to it 😃

0reactions

kstoneriv3commented, Apr 20, 2021

Is anyone working on this? I am interested in implementing this.

Top Results From Across the Web

sklearn.kernel_approximation.Nystroem

Approximate a kernel map using a subset of the training data. Constructs an approximate feature map for an arbitrary kernel using a subset...

sklearndf.transformation.NystroemDF — sklearndf documentation

inverse_transform. Inverse-transform the given inputs. set_params ... estimator ( Nystroem ) – the fitted native estimator to use as the delegate.

How to use scikit learn inverse_transform with new values

I know that StandardScaler.fit_transform() make the data have zero mean and unit variance. But with the new points of shape (num_clusters, ...

Fast brain decoding with random sampling and random ...

from sklearn.kernel_approximation import Nystroem class ... fit_transform(self, X, y=None): self.fit(X) return self.transform(X) def inverse_transform(self, ...

scikit-learn 1.0 Released - Exxact Corporation

Nystroem where the attribute component_indices_ did not correspond ... OrdinalEncoder.inverse_transform is not supporting sparse matrix and ...