Yuankai Luo (罗 元凯)

Email: luoyk@buaa.edu.cn, yuankluo@polyu.edu.hk

[Google Scholar] [Github]

I am a Ph.D. candidate in School of Computer Science and Engineering, Beihang University, supervised by Prof. Lei Shi and a joint Ph.D. student at The Hong Kong Polytechnic University under the supervision of Prof. Xiao-Ming Wu. Before that, I did research supervised by Veronika Thost. Previously, I received the BEng degree in Computer Science and Engineering from Chongqing University of Posts and Telecommunications, in 2021.

My research interests span Graph Neural Network (GNN) research, including architecture design, pre-training strategies, and compression/acceleration:

GNN Architecture and Analysis:

• Unified Framework GNN+ for Reassessing Classic GNNs in General Graph Tasks: developed GNN+, a framework that integrates message passing and well-known regularization techniques like dropout. GNN+ demonstrates that the true potential of classic GNNs has been previously underestimated in both node-level and graph-level tasks, challenging the belief that complex mechanisms are necessary for superior performance in graph models [NeurIPS 2024, arXiv 2025].
• Graph Transformers for Specialized Small-Scale Graph Tasks: designed specialized Graph Transformers architecture tailored for small-scale graph tasks, particularly directed acyclic graphs [NeurIPS 2023, KDD 2023], and graphs with multi-level structures [NeurIPS 2024].

GNN Pre-training: proposed a graph self-supervised learning framework based on persistent homology theory, which effectively captures the multi-scale topological features of graph data [NeurIPS 2023].

GNN Compression: introduced vector quantization to compress continuous node embeddings into highly compact (typically 6-15 dimensions), discrete (int4 type), and interpretable node representations—termed Node IDs [ICLR 2025].