Institution profile

University of Louisville

Academic institutionnorthamerica · us
Official website
Research library107linked papers
Opportunities0open roles
Selected work

Representative Papers

Do We Really Need to Design New Byzantine-robust Aggregation Rules?

Jan 29, 2025

In federated learning, Byzantine clients launching poisoning attacks can severely degrade the robustness of existing aggregation rules. To address this, we propose FoundationFL—a framework that preserves standard robust aggregators (e.g., Trimmed-mean, Median) without modifying their logic; instead, the server generates synthetic model updates, which are jointly aggregated with clients’ local updates. We provide the first theoretical proof that enhancing input quality alone—without designing new aggregation rules—significantly improves Byzantine resilience of classical robust aggregators. FoundationFL guarantees convergence under Byzantine threats and empirically demonstrates substantial improvements in poisoning resistance across multiple real-world datasets, while maintaining high model accuracy and low communication overhead. The framework thus achieves strong effectiveness, generalizability, and practicality.

1 citationsRead paper
Recent publications

Latest Papers

Separating Spatial and Clinical Risk with Node-Splitting SVM Survival Trees

Aug 13, 2026

This study addresses the challenge of disentangling spatial risk from clinical covariates in survival analysis by proposing a nonparametric two-stage residual separation framework. Leveraging kernel-bipolar splitting survival trees and cumulative hazard residuals, this method achieves risk decoupling without prespecified functional forms while establishing theoretical identifiability conditions. Experimental results demonstrate that the approach generates piecewise constant spatial risk maps with distinct boundaries, accurately identifying high-risk regions and sharp transitions in leukemia data. Furthermore, simulation studies confirm its robustness, effectively resolving confounding issues arising from spatially structured covariates. This work provides a rigorous solution for separating complex spatial effects from clinical features in survival modeling.

0 citationsRead paper