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.

Simplify spectral clustering solver logic

See original GitHub issue

We need to simplify the logic for selecting a solver in spectral_clustering. See discussion here: https://github.com/scikit-learn/scikit-learn/issues/10715#issuecomment-369236982 https://github.com/scikit-learn/scikit-learn/pull/14647#issuecomment-521294241 https://github.com/scikit-learn/scikit-learn/pull/10720#issuecomment-518878136

Issue Analytics

State:
Created 4 years ago
Comments:5 (5 by maintainers)

Top GitHub Comments

2reactions

alexshackedcommented, Oct 7, 2019

After we finish the refactoring and achieve a clear code, there are a couple of candidates for improving the algorithm. I found those by reading the issues referenced here. For now I have 2 ideas based on @lobpcg suggestions .

call scipy.linalg.eigh() always if n_nodes < 5 * n_components. Meaning, if number of samples is smaller than the (number_of_requested_eigen_vectors x 5) original comment: https://github.com/scikit-learn/scikit-learn/pull/14647#issuecomment-521076953
When calling arpack with shift-invert use a sigma value of -1e-5 instead of 1.0 and do not multiply the laplacian by -1, just before calling arpack original comment: https://github.com/scikit-learn/scikit-learn/pull/14647#issuecomment-521304431

@amueller what do you think?

0reactions

alexshackedcommented, Oct 4, 2019

After reading the issues and PRs referenced here and also studying the spectral clustering implementation, I understand that this enhancement’s goal is to refactor the logic for chosing an eigen solver in function spectral_embedding() located in manifold/spectral_embedding.py I opened a [WIP ] PR #15136, where I show an inital suggestion for refactoring spectral_embedding(). My intention is just to present an idea for simplifing the logic that choses the solver. It felt easier to write code, then to write a design document. I hope to get feed-back that will focus me onwards. I also intend to add more regressinon tests, but I thought to work on the tests only after the refactoring goal becomes more clear to me.

Top Results From Across the Web

A Tutorial on Spectral Clustering - People

It is simple to implement, can be solved efficiently by standard linear algebra software, and very often outperforms traditional clustering algorithms such as ......

A Tutorial on Spectral Clustering

It is simple to implement, can be solved efficiently by standard linear algebra software, and very often outperforms traditional clustering algorithms such as ......

Understanding Spectral Clustering - Eran Raviv

Spectral clustering includes a processing step to help solve non-linear problems, such that they ... A typical spectral clustering algorithm.

Self-constrained Spectral Clustering | IEEE Journals & Magazine

To solve this problem, we propose a self-constrained spectral clustering algorithm. In this algorithm, we extend the objective function of ...

Foundations of a Multi-way Spectral Clustering Framework for ...

Different strategies have been proposed to solve this problem, however, ... the Theoretical Spectral Curvature Clustering (TSCC) algorithm for solving the ...