AI Paper Atlas

Research map / Federated learning and privacy / Privacy attacks and secure learning

Robust and Actively Secure Serverless Collaborative Learning

Nicholas Franzese, Adam Dziedzic, Christopher A. Choquette-Choo, Mark R. Thomas, Muhammad Ahmad Kaleem, Stephan Rabanser, Congyu Fang, Somesh Jha, Nicolas Papernot, Xiao Wang

NeurIPS 2023 · Federated learning and privacy › Privacy attacks and secure learning · approach: Membership inference and auditing · 2 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Use malicious-secure P2P RSA aggregationP2P collaborative learning · Aggregation step, 1000 peers, 10^5 parametersCPU time: 46 secondsImproved
Represent updates as bits using RSA rather than fixed-point updates with Centered ClippingP2P aggregation · RSA vs. Centered Clippingaggregation speed: around ~30XImproved
Represent updates as bits using RSA rather than fixed-point updates with FLTrustP2P aggregation · RSA vs. FLTrustaggregation speed: around ~6XImproved
Use malicious-secure Centered Clipping and FLTrust instead of RSAP2P learning · EMNISTconvergence speedImproved
Use Centered Clipping or FLTrust instead of RSAP2P learning · EMNISTfinal accuracyImproved
Use Centered Clipping instead of FLTrustP2P learning · EMNISTaccuracyImproved
Modify Centered Clipping from ℓ2 clipping to ℓ∞ box clipping (ablation)Centered Clipping · IID EMNIST and MNIST under attacksrobustnessNo clear effect
Lower Centered Clipping numerical precision to fixed-point (ablation)Centered Clipping · IID MNIST under ALIE attackrobustnessNo clear effect
Use the modified malicious-secure versions of CC, FLTrust, and RSA (ablation)Robust aggregation algorithms · Robustness evaluationsmodel performanceNo clear effect
Reduce FLTrust fixed-point length from 32 bits to 16 bits relative to CCFLTrust · P2P trainingcomputational efficiency: 16 bits vs 32 bitsImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy