Research map / Federated learning and privacy / Communication-efficient FL
TRAIL: Trust-Aware Client Scheduling for Semi-Decentralized Federated Learning
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes TRAIL, a trust-aware client scheduling mechanism for semi-decentralized federated learning that uses an adaptive hidden semi-Markov model to estimate client trust and a greedy algorithm to assign clients to edge servers.
Problem
Semi-decentralized federated learning suffers from dynamic, unreliable clients (fluctuating training quality, packet loss, intermittent participation), and existing methods do not jointly handle client-state prediction and client selection.
Approach
Combine an adaptive hidden semi-Markov model that predicts each client's communication/training quality and a trust level with a client-server association optimization solved by a greedy algorithm derived from a convergence bound.
Key results
- TRAIL reports an 8.7% improvement in test accuracy and a 15.3% reduction in training loss over baselines.
- Experiments use 5 edge servers and 50 clients, with 10%, 30%, and 50% low-quality clients.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Combine AHSMM client prediction with greedy client scheduling | TRAIL · MNIST, EMNIST, CIFAR10, SVHN | test accuracy: 8.7% | Helped |
| Combine AHSMM client prediction with greedy client scheduling | TRAIL · MNIST, EMNIST, CIFAR10, SVHN | training loss: 15.3% | Helped |
Limitations
- Convergence analysis relies on strong convexity and Lipschitz gradient assumptions, while experiments use CNNs in non-convex settings.
- Reported gains are relative to only three baselines, and the 8.7% and 15.3% figures are not broken down per dataset in the text.
Experimental setup
Datasets: MNIST EMNIST CIFAR-10 SVHN
Benchmarks: MNIST EMNIST CIFAR-10 SVHN
Metrics: test accuracy training loss
Baselines: GUROBI TACS (TRUST) RANDOM
Builds on: hidden semi-Markov model Viterbi algorithm Baum-Welch algorithm FedAvg
Closest papers by meaning
- Low Precision Local Training is Enough for Federated Learning (NeurIPS 2024 · similarity 0.89)
- Federated Learning under Periodic Client Participation and Heterogeneous Data: A New Communication-Efficient Algorithm and Analysis (NeurIPS 2024 · similarity 0.88)
- SemiDFL: A Semi-Supervised Paradigm for Decentralized Federated Learning (AAAI 2025 · similarity 0.88)
- FedCompass: Efficient Cross-Silo Federated Learning on Heterogeneous Client Devices Using a Computing Power-Aware Scheduler (ICLR 2024 · similarity 0.88)
- Byzantine-robust Decentralized Federated Learning via Dual-domain Clustering and Trust Bootstrapping (CVPR 2024 · similarity 0.88)
- FedLMT: Tackling System Heterogeneity of Federated Learning via Low-Rank Model Training with Theoretical Guarantees (ICML 2024 · similarity 0.88)
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.