12 papers
FocusOPD: Structure-Guided On-Policy Distillation for Diffusion Models
Xinrui Lin, Shumin Wang, Suqi Song +2
Multi-teacher on-policy distillation enables rectified-flow models with diffusion-transformer architectures to combine task-specific capabilities within a single student. However,…
When Pairwise Preferences Cycle: A Reliability-Anchored Preference Walk for Video GRPO
Xinrui Lin, Fan Li, Shumin Wang +2
Pairwise preference feedback has become a widely adopted supervision signal for reinforcement learning in video generation models, as pairwise comparisons provide a more reliable a…
Answer-then-Edit: Reasoning Skeleton Editing for Anti-Distillation with Preserved Utility
Fan Li, Mengting Pan, Sijia Xu +3
Proprietary large language models (LLMs) entail substantial intellectual and financial investment, making them valuable intellectual property (IP). However, even when deployed via…
Universal Graph Backdoor Defense: A Feature-based Homophily Perspective
Mengting Pan, Fan Li, Chen Chen +1
Graph neural networks (GNNs) have achieved remarkable success in relational learning. However, their vulnerability to graph backdoor attacks (GBAs) poses a significant barrier to b…
Anchor-guided Hypergraph Condensation with Dual-level Discrimination
Fan Li, Xiaoyang Wang, Chen Chen +1
The increasing prevalence of large-scale hypergraphs poses significant computational challenges for hypergraph neural network (HNN) training. To address this, hypergraph condensati…
CTC: A Training-Free Framework for Efficient Tabular Data Condensation
Sijia Xu, Fan Li, Xiaoyang Wang +2
Tabular data is the primary data format in industrial relational databases, underpinning modern data analytics and decision-making. However, the increasing scale of tabular data po…
DHG-Bench: A Comprehensive Benchmark for Deep Hypergraph Learning
Fan Li, Xiaoyang Wang, Wenjie Zhang +2
Deep graph models have achieved great success in network representation learning. However, their focus on pairwise relationships restricts their ability to learn pervasive higher-o…
Graph is a Natural Regularization: Revisiting Vector Quantization for Graph Representation Learning
Zian Zhai, Fan Li, Xingyu Tan +2
Vector Quantization (VQ) has recently emerged as a promising approach for learning compressed and discrete representations for graph-structured data. However, a fundamental challen…
HiTeC: Hierarchical Contrastive Learning on Text-Attributed Hypergraph with Semantic-Aware Augmentation
Mengting Pan, Fan Li, Chen Chen +2
Contrastive learning (CL) has become a dominant paradigm for self-supervised hypergraph learning, enabling effective training without costly labels. However, node entities in real-…
Efficient Dynamic Attributed Graph Generation
Fan Li, Xiaoyang Wang, Dawei Cheng +3
Data generation is a fundamental research problem in data management due to its diverse use cases, ranging from testing database engines to data-specific applications. However, rea…
TCGU: Data-centric Graph Unlearning based on Transferable Condensation
Fan Li, Xiaoyang Wang, Dawei Cheng +3
With growing demands for data privacy and model robustness, graph unlearning (GU), which erases the influence of specific data on trained GNN models, has gained significant attenti…
Hypergraph Self-supervised Learning with Sampling-efficient Signals
Fan Li, Xiaoyang Wang, Dawei Cheng +3
Self-supervised learning (SSL) provides a promising alternative for representation learning on hypergraphs without costly labels. However, existing hypergraph SSL models are mostly…