Full Professor, National University of Defense Technology
6 papers at NeurIPS 2025
The paper systematically analyzes the phenomenon of varying sparsity ratios across views in multi-view learning, and proposed a targeted data-driven network architecture based on Sparse Autoencoder with Adaptive Constraints.
We propose the first comprehensive and unified benchmark for deep graph clustering, offer an open-source package named PyDGC, and point out promising research directions for DGC from extensive experimental analyses.
