search

yzhao062 / pyod

A Python Library for Outlier and Anomaly Detection, Integrating Classical and Deep Learning Techniques
http://pyod.readthedocs.io
BSD 2-Clause "Simplified" License
8.46k stars 1.36k forks source link

Support for kernel PCA method to handle nonlinear data distribution #370

Open highbo opened 2 years ago

highbo commented 2 years ago

Is it planned to implement kernel PCA method in PYOD?

Reference: Section 3.3.8 (, and Section 4.2.1) of "Outlier Analysis" by C.C. Aggarwal

tam17aki commented 2 years ago

Hi,

Recently I have tried to implement the method based on kernel PCA, according to the PyOD format.

Class module: https://gist.github.com/tam17aki/9bd7d48c11ece38df8e086ab3d5f143d

Example: https://gist.github.com/tam17aki/03815ba272c52d39eb4b577105e1a10e

Benchmarks: https://gist.github.com/tam17aki/03ebe0f7dbd02364a49211f6855fa88b

In future, I would like to add the module to PyOD.

Best regards, Akira

tam17aki commented 2 years ago

Hi,

I have created the pull-req for kernel PCA. https://github.com/yzhao062/pyod/pull/438

I hope this will be merged in near future.

Best regards, Akira

highbo commented 2 years ago

Hi Akira,

Thank you for your update! This is great! I will check it out. Hope to see it merged soon.

Best, Houwu

On Wed, Sep 14, 2022 at 1:27 AM Akira Tamamori @.***> wrote:

Hi,

I have created the pull-req for kernel PCA.

438 https://github.com/yzhao062/pyod/pull/438

I hope this will be merge in near future.

Best regards, Akira

— Reply to this email directly, view it on GitHub https://github.com/yzhao062/pyod/issues/370#issuecomment-1246419684, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACYNEHEIJXCYXZOWRBG2G6TV6GD5XANCNFSM5OY27Y4Q . You are receiving this because you authored the thread.Message ID: @.***>