Ping Xiong

Deep GNNs can achieve superior depth scaling and accuracy by leveraging controlled Turing instability or stable near-critical propagation to combat over-smoothing.

Explicitly modeling the temporal rate of change in spatio-temporal graph models unlocks superior prediction accuracy and pattern preservation for propagating phenomena.

RISA achieves a notable improvement in refusal reliability for LLMs while preserving their response utility, transforming how we approach inference-time alignment.

A new attribution method reveals the full spectrum of information flow in ETGNNs, enhancing explainability beyond traditional event-related embeddings.

Exponential complexity in GNN explainability is no longer a barrier: this work achieves linear-time subgraph attribution via message passing.