Jian Li

collaborators

2 papers

cs.LG2026

PID-Guided Partial Alignment for Multimodal Decentralized Federated Learning

Yanhang Shi, Xiaoyu Wang, Houwei Cao +2

Multimodal decentralized federated learning (DFL) must support collaboration among agents that hold different modality subsets and often different model components, while operating…

cs.LG2026

When Does $ℓ_{2}$ -Boosting Overfit Benignly? High-Dimensional Risk Asymptotics and the $ℓ_{1}$ Implicit Bias

Ye Su, Jian Li, Yong Liu

Benign overfitting is well-characterized in $ℓ_{2}$ geometries, but its behavior under the $ℓ_{1}$ implicit bias of greedy ensembles remains challenging. The analytical barrier s…

PID-Guided Partial Alignment for Multimodal Decentralized Federated Learning

When Does ℓ2​-Boosting Overfit Benignly? High-Dimensional Risk Asymptotics and the ℓ1​ Implicit Bias

When Does $ℓ_{2}$ -Boosting Overfit Benignly? High-Dimensional Risk Asymptotics and the $ℓ_{1}$ Implicit Bias