NeurIPS22-FMVACC

(NeurIPS22) Align then Fusion: Generalized Large-scale Multi-view Clustering with Anchor Matching Correspondences

Multi-view anchor graph clustering

For seeking efficiency for graph methods, anchor-based (landmark/prototype) method has become a popular choice. For multi-view setting, anchors should be seriously considered with alignment issue then multi-view fusion can be correctly done.

Simulated datasets

Dataset Preparation：

datasets/synthetic_data. You can also generate your own multi-view data to see Anchor-unaligned Problem (Dimension 2)

Generate anchors:

You can choose sampling/$k$-means, or the learned anchors by our paper.

AUP problems:

If the anchors are generated independently on each view, then cross-view anchor alignment is needed. Sometimes, the anchors may be exactly aligned with nice initializations.

Real multi-view datasets

Run

run ours_aligned/run.m