Research map / Federated learning and privacy / Communication-efficient FL
Memory-Based Optimization Methods for Model-Agnostic Meta-Learning and Personalized Federated Learning
Bokun Wang, Zhuoning Yuan, Yiming Ying, Tianbao Yang
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 |
|---|---|---|---|
| add MOML-v1 memory-based update instead of MAML (BSGD) | MAML · Sinewave regression (K=1) | avg test error: 0.889 -> 0.291 | Improved |
| use MOML-v1 instead of MAML (BSGD) | MAML · Sinewave regression (K=3) | avg test error: 0.321 -> 0.196 | Improved |
| use MOML-v2 instead of MOML-v1 (update all n personalized models) | MOML · Sinewave regression (K=1) | avg test error: 0.291 -> 0.448 | Hurt |
| use MOML-v2 instead of MOML-v1 | MOML · Sinewave regression (K=1) | avg time per iteration (ms): 2.180 -> 7.863 | Hurt |
| use LocalMOML instead of MOML-v1 | MOML · Sinewave regression (K=3) | avg test error: 0.196 -> 0.170 | Improved |
| use MOML-v1 instead of MAML (BSGD) | MAML · Omniglot 20-way one-shot | test accuracy (%): 44.31 -> 46.35 | Improved |
| use MOML-v2 instead of MAML (BSGD) | MAML · CIFAR-100 5-way one-shot | test accuracy (%): 40.18 -> 40.82 | Improved |
| use LocalMOML instead of Per-FedAvg | Per-FedAvg · CIFAR-100 federated 5-shot (H=10, B'=5 tasks per GPU) | test accuracy (%): 53.37 -> 56.45 | Improved |
| use LocalMOML instead of pFedMe | pFedMe · CIFAR-100 federated 5-shot (H=4, B'=1 tasks per GPU) | test accuracy (%): 49.64 -> 54.00 | Improved |
| use LocalMOML instead of Per-FedAvg | Per-FedAvg · CIFAR-10 federated 5-shot (H=4, B'=1 tasks per GPU) | test accuracy (%): 63.81 -> 66.16 | Improved |
| use LocalMOML instead of Per-FedAvg | Per-FedAvg · MNIST federated 5-shot (H=4, B'=1 tasks per GPU) | test accuracy (%): 91.63 -> 91.62 | No clear effect |
| use ProtoNet instead of MAML (BSGD) | MAML · CIFAR-100 5-way one-shot | test accuracy (%): 40.18 -> 31.11 | Hurt |
Closest papers by meaning
- Personalized Federated Learning with Theoretical Guarantees: A Model-Agnostic Meta-Learning Approach (NeurIPS 2020 · similarity 0.90)
- On the Global Optimality of Model-Agnostic Meta-Learning (ICML 2020 · similarity 0.88)
- Generalization of Model-Agnostic Meta-Learning Algorithms: Recurring and Unseen Tasks (NeurIPS 2021 · similarity 0.88)
- Memory-Reduced Meta-Learning with Guaranteed Convergence (AAAI 2025 · similarity 0.86)
- FedQS: Optimizing Gradient and Model Aggregation for Semi-Asynchronous Federated Learning (NeurIPS 2025 · similarity 0.85)
- Exact and Linear Convergence for Federated Learning under Arbitrary Client Participation is Attainable (NeurIPS 2025 · similarity 0.85)
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.
