mlsanityData to 14 Oct 2026

Generalization of Model-Agnostic Meta-Learning Algorithms: Recurring and Unseen Tasks

arXiv 2102.03832 ↗First posted 7 Feb 2021cs.LG math.OC stat.MLcited by 36

Not on the map. Why?

Abstract (arXiv)

In this paper, we study the generalization properties of Model-Agnostic Meta-Learning (MAML) algorithms for supervised learning problems. We focus on the setting in which we train the MAML model over m tasks, each with n data points, and characterize its generalization error from two points of view: First, we assume the new task at test time is one of the training tasks, and we show that, for strongly convex objective functions, the expected excess population loss is bounded by O(1/mn). Second, we consider the MAML algorithm's generalization to an unseen task and show that the resulting generalization error depends on the total variation distance between the underlying distributions of the new task and the tasks observed during the training process. Our proof techniques rely on the connections between algorithmic stability and generalization bounds of algorithms. In particular, we propose a new definition of stability for meta-learning algorithms, which allows us to capture the role of both the number of tasks m and number of samples per task n on the generalization error of MAML.

Key facts. Cited by 36 papers in the map (data through 14 Oct 2026). Builds on 22 of its 36 references that are in the map. First posted on arXiv 7 Feb 2021.

Builds on

22 of 36 references.

References · 22

  1. 2017Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networkscited by 4,897
  2. 2016Neural Architecture Search with Reinforcement Learningcited by 1,499
  3. 2017Learning Transferable Architectures for Scalable Image Recognitioncited by 1,101
  4. 2018On First-Order Meta-Learning Algorithmscited by 1,018
  5. 2016Learning to learn by gradient descent by gradient descentcited by 620
  6. 2015Train faster, generalize better: Stability of stochastic gradient descentcited by 552
  7. 2019Meta-Learning with Implicit Gradientscited by 525
  8. 2017Meta-SGD: Learning to Learn Quickly for Few-Shot Learningcited by 379
  9. 2018How to train your MAMLcited by 309
  10. 2016Designing Neural Network Architectures using Reinforcement Learningcited by 279
  11. 2018Meta-Learning: A Surveycited by 256
  12. 2019Private Stochastic Convex Optimization with Optimal Ratescited by 186
10 more
  1. 2019Adaptive Gradient-Based Meta-Learning Methodscited by 183
  2. 2018Recasting Gradient-Based Meta-Learning as Hierarchical Bayescited by 180
  3. 2019On the Convergence Theory of Gradient-Based Model-Agnostic Meta-Learning Algorithmscited by 132
  4. 2020Solving Stochastic Compositional Optimization is Nearly as Easy as Solving Stochastic Optimizationcited by 51
  5. 2020Theoretical Convergence of Multi-Step Model-Agnostic Meta-Learningcited by 28
  6. 2019Alpha MAML: Adaptive Model-Agnostic Meta-Learningcited by 25
  7. 2020On the Global Optimality of Model-Agnostic Meta-Learningcited by 21
  8. 2019Towards Understanding Generalization in Gradient-Based Meta-Learningcited by 12
  9. 2020Meta Learning in the Continuous Time Limitcited by 10
  10. 2020UFO-BLO: Unbiased First-Order Bilevel Optimizationcited by 3

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2021: 5; 2022: 8; 2023: 9; 2024: 4; 2025: 6; 2026: 4.

36 papers.

Sort:

By area:

  1. 2021Towards Gradient-based Bilevel Optimization with Non-convex Followers and Beyondcited by 63
  2. 2023Understanding Generalization of Federated Learning via Stability: Heterogeneity Matterscited by 27
  3. 2023Federated Domain Generalization: A Surveycited by 25
  4. 2025Towards Agentic AI Networking in 6G: A Generative Foundation Model-as-Agent Approachcited by 19
  5. 2022Is Bayesian Model-Agnostic Meta Learning Better than Model-Agnostic Meta Learning, Provably?cited by 13
  6. 2021FLIX: A Simple and Communication-Efficient Alternative to Local Methods in Federated Learningcited by 12
  7. 2022FORML: Learning to Reweight Data for Fairnesscited by 12
  8. 2022Understanding Benign Overfitting in Gradient-Based Meta Learningcited by 11
  9. 2022Distributionally Adaptive Meta Reinforcement Learningcited by 11
  10. 2023A Simple Yet Effective Strategy to Robustify the Meta Learning Paradigmcited by 10
  11. 2021Two Sides of Meta-Learning Evaluation: In vs. Out of Distributioncited by 9
  12. 2021Learning an Explicit Hyperparameter Prediction Function Conditioned on Taskscited by 9
24 more
  1. 2022Provable Generalization of Overparameterized Meta-learning Trained with SGDcited by 9
  2. 2023Mnemosyne: Learning to Train Transformers with Transformerscited by 9
  3. 2022A Projection-free Algorithm for Constrained Stochastic Multi-level Composition Optimizationcited by 8
  4. 2023Bayes meets Bernstein at the Meta Level: an Analysis of Fast Rates in Meta-Learning with PAC-Bayescited by 8
  5. 2021Uniform Sampling over Episode Difficultycited by 6
  6. 2022Log-Euclidean Signatures for Intrinsic Distances Between Unaligned Datasetscited by 5
  7. 2022PersA-FL: Personalized Asynchronous Federated Learningcited by 5
  8. 2023RotoGBML: Towards Out-of-Distribution Generalization for Gradient-Based Meta-Learningcited by 5
  9. 2023M-L2O: Towards Generalizable Learning-to-Optimize by Test-Time Fast Self-Adaptationcited by 4
  10. 2025Is Meta-Learning Out? Rethinking Unsupervised Few-Shot Classification with Limited Entropycited by 4
  11. 2024Meta-Reinforcement Learning with Universal Policy Adaptation: Provable Near-Optimality under All-task Optimum Comparatorcited by 2
  12. 2025An Information-Theoretic Analysis for Federated Learning under Concept Driftcited by 2
  13. 2023Zero-Shot Anomaly Detection via Batch Normalizationcited by 1
  14. 2024Rethinking Meta-Learning from a Learning Lenscited by 1
  15. 2024Exploring the Generalization Capabilities of AID-based Bi-level Optimizationcited by 1
  16. 2026Stability beyond Bounded Differences: Sharp Generalization Bounds under Finite Lp Momentscited by 1
  17. 2023First-order ANIL provably learns representations despite overparametrizationcited by 0
  18. 2024Bayesian meta learning for trustworthy uncertainty quantificationcited by 0
  19. 2025Enhancing Environmental Robustness in Few-shot Learning via Conditional Representation Learningcited by 0
  20. 2025Revisiting Meta-Learning with Noisy Labels: Reweighting Dynamics and Theoretical Guaranteescited by 0
  21. 2025An Information-Theoretic Analysis of OOD Generalization in Meta-Reinforcement Learningcited by 0
  22. 2026Bayesian Meta-Learning with Expert Feedback for Task-Shift Adaptation through Causal Embeddingscited by 0
  23. 2026RELOAD: A Robust and Efficient Learned Query Optimizer for Database Systemscited by 0
  24. 2026Machine Learning for Underwater Optical Wireless Communication Systems: A Comprehensive Surveycited by 0