James Sudlow (PhD candidate), Baudouin Naline (post-doctoral researcher), and Sara Bouchenak receive the Distinguished Artifact Award of the IEEE/IFIP DSN Conference 2026.

And congratulations for James for his (first) presentation at DSN!

What is the paper about?

Data cleaning is often treated as a harmless preprocessing step in Federated Learning—but is it really? Our large-scale empirical study reveals that it can significantly reshape both model performance and fairness, raising important questions for trustworthy FL.

The code is available here: https://lnkd.in/e36GWBB5

And the paper will soon appear in the conference proceedings.