Research map / Federated learning and privacy / Privacy attacks and secure learning
Bant: Byzantine Antidote via Trial Function and Trust Scores
Gleb Molodtsov, Daniil Medyakov, Sergey Skorik, Nikolas Khachaturov, Shahane Tigranyan, Vladimir Aletov, Aram Avetisyan, Martin Takáč, Aleksandr Beznosikov
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 changed | Where | Measured | Effect |
|---|---|---|---|
| use trial-function trust scores (Bant) instead of Adam averaging | Adam · ECG AFIB, Label Flipping (60%) | G-mean: 0.262 -> 0.956 | Improved |
| use Bant instead of FLTrust under Random Gradients majority | FLTrust · ECG AFIB, Random Gradients (60%) | G-mean: 0.617 -> 0.948 | Improved |
| use Bant instead of SafeGuard under IPM 80% | SafeGuard · ECG AFIB, IPM (80%) | f1-score: 0.082 -> 0.676 | Improved |
| add Bant trust scores instead of Centered Clip on ALIE | Centered Clip · ECG AFIB, ALIE (40%) | G-mean: 0.530 -> 0.947 | Improved |
| add fixing-by-mixing to Centered Clip | Centered Clip · ECG AFIB, ALIE (40%) | G-mean: 0.530 -> 0.876 | Improved |
| add fixing-by-mixing to Centered Clip | Centered Clip · ECG AFIB, IPM (80%) | G-mean: 0.084 -> 0.027 | Hurt |
| use auxiliary mirror-descent weight optimization (AutoBant) instead of Bant | Bant · ECG AFIB, ALIE (40%) | G-mean: 0.947 -> 0.892 | Hurt |
| use AutoBant instead of Bant under 60% label flipping | Bant · ECG AFIB, Label Flipping (60%) | f1-score: 0.777 -> 0.276 | Hurt |
| use Bant under no attack versus Adam | Adam · ECG AFIB, no attack | G-mean: 0.956 -> 0.953 | No clear effect |
| use VR MARINA variance reduction under no attack | Adam · ECG AFIB, no attack | G-mean: 0.956 -> 0.010 | Hurt |
Closest papers by meaning
- From Inexact Gradients to Byzantine Robustness: Acceleration and Optimization under Similarity (NeurIPS 2026 · similarity 0.85)
- On the Tension between Byzantine Robustness and No-Attack Accuracy in Distributed Learning (ICML 2025 · similarity 0.84)
- Weight for Robustness: A Comprehensive Approach towards Optimal Fault-Tolerant Asynchronous ML (NeurIPS 2024 · similarity 0.84)
- Dynamic Byzantine-Robust Learning: Adapting to Switching Byzantine Workers (ICML 2024 · similarity 0.84)
- Byzantine-Robust Decentralized Learning via Remove-then-Clip Aggregation (AAAI 2024 · similarity 0.84)
- ByMI: Byzantine Machine Identification with False Discovery Rate Control (ICML 2024 · similarity 0.83)
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.
