We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Machine Learning

Title: Generalization Bounds For Meta-Learning: An Information-Theoretic Analysis

Abstract: We derive a novel information-theoretic analysis of the generalization property of meta-learning algorithms. Concretely, our analysis proposes a generic understanding of both the conventional learning-to-learn framework and the modern model-agnostic meta-learning (MAML) algorithms. Moreover, we provide a data-dependent generalization bound for a stochastic variant of MAML, which is non-vacuous for deep few-shot learning. As compared to previous bounds that depend on the square norm of gradients, empirical validations on both simulated data and a well-known few-shot benchmark show that our bound is orders of magnitude tighter in most situations.
Subjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
Cite as: arXiv:2109.14595 [cs.LG]
  (or arXiv:2109.14595v2 [cs.LG] for this version)

Submission history

From: Qi Chen [view email]
[v1] Wed, 29 Sep 2021 17:45:54 GMT (463kb,D)
[v2] Fri, 10 Dec 2021 16:31:27 GMT (470kb,D)

Link back to: arXiv, form interface, contact.