Hi! Is there any python implementation available for Incremental KPCA? I checked the only available git-hub repository (https://github.com/fredhallgren/inkpca) for incremental KPCA. However, it does not work for very high dimensional data, for example, 3100 features. Is there anyone who has used this repository for very high dimensional data or can anyone provide any explanation for this repository for not working on high dimensional data?
Hi, Check the following links:
https://github.com/fredhallgren/inkpca
https://scikit-learn.org/stable/modules/generated/sklearn.decomposition.IncrementalPCA.html#sklearn.decomposition.IncrementalPCA
Hi Hamid Nasiri , I already checked the git hub link that you provided. This repository does not work for really high-dimensional data, as I mentioned in my question. Also, the scikit learn link is for incremental PCA not incremental KPCA.
Hi Farzana Anowar , I think KPCA in general has a problem with large feature numbers. See Chin et al 2007: "However, the standard implementation of KPCA scales badly with the problem size, making computations for large problems infeasible."
Article Incremental Kernel Principal Component Analysis
So I am not sure if any other package will solve your problem, as the problem seems to be part of the KPCA algorithm itself. What exactly are you trying to do? Could other clustering methods work for you as well? What about t-SNE?
Hi Hannes Reinwald , the regular KPCA performs fine with this dataset that I am trying with the incremental KPCA.
Try sklearn.decomposition.KernelPCA
https://scikit-learn.org/stable/modules/generated/sklearn.decomposition.KernelPCA.html
@Anuraj Nayarisseri, I'm looking for Incremental KPCA, not the regular one.
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Programming Languages
- Python