Publications
PosterCopilot: Toward Layout Reasoning and Controllable Editing for Professional Graphic Design
arXiv, 2025
, Ken Li*, Tianyu Lao, Haofan Wang, Liang Wang, Caifeng Shan, Chenyang Si
[PDF] [Project Page] [Code]
TL;DR: We introduce PosterCopilot, the first graphic design framework that fully integrates visual feedback with human aesthetics. Leveraging our Generative Agent and the design model’s powerful reasoning capabilities, it enables layer-level, multi-round, high-fidelity editing.
ALM-PU: Positive and Unlabeled learning with constrained optimization
ECML-PKDD, 2025
, Yuefei Wu*, Ken Li, Bo Dong, Bin Shi†
[PDF] [Code]
TL;DR: We propose a novel constraint optimization-based PU-learning framework, which enhances sample discrimination capability by imposing distinct mathematical constraints on samples and resolving the problem through systematic optimization.
NI-GDBA: Non-Intrusive Distributed Backdoor Attack Based on Adaptive Perturbation on Federated Graph Learning
WWW, 2025
Ken Li, Bin Shi, , Bo Dong
[PDF]
TL;DR: We propose a non-intrusive graph distributed backdoor attack (NI-GDBA) that does not require backdoor triggers to be injected in the training data.
* Equal contribution. † Corresponding author.