This is the experiment code of https://arxiv.org/abs/2211.15953, a Decentralized algorithm for Kernel Principal Component Analysis (DeKPCA) for sample-distributed setting, where each local agent contains a subset of samples with full features.
This is MatLab code of our paper, which is used to test the accuracy of the result of our algorithm.
To see the truly parallel implementation, please refer in the Python code: https://github.com/hefansjtu/DKPCA-ADMM
For test, please run ''test.m'' directly.
Eigenvectors computed by SVD is regarded as ground truth.