Synthetic dataset for personalized federated unsupervised learning

The dataset used in the paper is a synthetic dataset for personalized federated unsupervised learning. It is generated by sampling a global PC, V ∗ ∈ St(d, r), uniformly on the Steifel manifold. The dataset is used to evaluate the performance of the proposed algorithms for personalized adaptive dimensionality reduction and personalized federated diffusion models.