mlsanityData to 7 Oct 2026

On Enforcing Better Conditioned Meta-Learning for Rapid Few-Shot Adaptation

arXiv 2206.07260 ↗First posted 15 Jun 2022cs.LG cs.CVcited by 7

Not on the map. Why?

Abstract (arXiv)

Inspired by the concept of preconditioning, we propose a novel method to increase adaptation speed for gradient-based meta-learning methods without incurring extra parameters. We demonstrate that recasting the optimization problem to a non-linear least-squares formulation provides a principled way to actively enforce a textitwell-conditioned parameter space for meta-learning models based on the concepts of the condition number and local curvature. Our comprehensive evaluations show that the proposed method significantly outperforms its unconstrained counterpart especially during initial adaptation steps, while achieving comparable or better overall results on several few-shot classification tasks -- creating the possibility of dynamically choosing the number of adaptation steps at inference time.

Key facts. Cited by 7 papers in the map (data through 7 Oct 2026). Builds on 25 of its 44 references that are in the map. First posted on arXiv 15 Jun 2022.

Builds on

25 of 44 references.

References · 25

  1. 2014ImageNet Large Scale Visual Recognition Challengecited by 8,441
  2. 2017Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networkscited by 4,861
  3. 2016Matching Networks for One Shot Learningcited by 2,246
  4. 2018On First-Order Meta-Learning Algorithmscited by 1,003
  5. 2019A Closer Look at Few-shot Classificationcited by 679
  6. 2016Learning to learn by gradient descent by gradient descentcited by 616
  7. 2019Meta-Learning with Implicit Gradientscited by 523
  8. 2018Meta-Learning with Latent Embedding Optimizationcited by 476
  9. 2018Meta-Learning for Semi-Supervised Few-Shot Classificationcited by 456
  10. 2018Shampoo: Preconditioned Stochastic Tensor Optimizationcited by 413
  11. 2018TADAM: Task dependent adaptive metric for improved few-shot learningcited by 408
  12. 2018Meta-learning with differentiable closed-form solverscited by 395
13 more
  1. 2017Meta-SGD: Learning to Learn Quickly for Few-Shot Learningcited by 377
  2. 2019Rapid Learning or Feature Reuse? Towards Understanding the Effectiveness of MAMLcited by 359
  3. 2018How to train your MAMLcited by 306
  4. 2018Meta-Transfer Learning for Few-Shot Learningcited by 306
  5. 2018Probabilistic Model-Agnostic Meta-Learningcited by 254
  6. 2018Recasting Gradient-Based Meta-Learning as Hierarchical Bayescited by 180
  7. 2017Continuous Adaptation via Meta-Learning in Nonstationary and Competitive Environmentscited by 106
  8. 2019Meta-Learning with Warped Gradient Descentcited by 98
  9. 2019Meta-Curvaturecited by 70
  10. 2019Provable Guarantees for Gradient-Based Meta-Learningcited by 67
  11. 2021Reproducibility Report: La-MAML: Look-ahead Meta Learning for Continual Learningcited by 42
  12. 2018Few-Shot Learning with Metric-Agnostic Conditional Embeddingscited by 40
  13. 2018Deep Meta-Learning: Learning to Learn in the Concept Spacecited by 22

Built on by

Cited by year

Citing papers per year: 2022: 1; 2023: 4; 2024: 1; 2025: 1.

7 papers.

  1. 2023Advances and Challenges in Meta-Learning: A Technical Reviewcited by 65
  2. 2022Gradient-based Bi-level Optimization for Deep Learning: A Surveycited by 26
  3. 2023Meta-Learning with a Geometry-Adaptive Preconditionercited by 11
  4. 2023SHOT: Suppressing the Hessian along the Optimization Trajectory for Gradient-Based Meta-Learningcited by 1
  5. 2023Meta-Learning with Versatile Loss Geometries for Fast Adaptation Using Mirror Descentcited by 1
  6. 2024Memory-Reduced Meta-Learning with Guaranteed Convergencecited by 0
  7. 2025Learnable Loss Geometries with Mirror Descent for Scalable and Convergent Meta-Learningcited by 0