Built independently by an author, for readers. Read the story and support ChapterPal
ChapterPal Logo
Log inSign up

keyword

few-shot learning performance

Few-shot learning performance measures how accurately and effectively a machine learning model generalizes to new, previously unseen tasks or classes when given only a very small number of labeled training examples. In meta-learning and sample-efficient artificial intelligence frameworks, it evaluates a system's ability to leverage prior experience across related tasks to quickly adapt parameters or representations using minimal data. Typically assessed on standardized benchmarks through metrics such as classification accuracy or loss on an evaluation set after adapting to a limited support set, high few-shot learning performance reflects robust task-level generalization, rapid convergence, and data efficiency without suffering from severe overfitting.

1 item

On First-Order Meta-Learning Algorithms

On First-Order Meta-Learning Algorithms

Alex Nichol, Joshua Achiam, John Schulman

OrganizationsOpenAI

Why you should read this

Introduces Reptile, a computationally efficient first-order meta-learning algorithm that avoids expensive second-order derivatives in MAML while matching its few-shot classification performance, supported by theoretical analysis explaining how first-order updates find effective initializations across tasks.

This paper considers meta-learning problems, where there is a distribution of tasks, and we would like to obtain an agent that performs well (i.e., learns quickly) when presented with a previously unseen task sampled from this distribution. We analyze a family of algorithms for learning a parameter initialization that can be fine-tuned quickly on a new task, using only first-order derivatives for the meta-learning updates. This family includes and generalizes first-order MAML, an approximation to MAML obtained by ignoring second-order derivatives. It also includes Reptile, a new algorithm that we introduce here, which works by repeatedly sampling a task, training on it, and moving the initialization towards the trained weights on that task. We expand on the results from Finn et al. showing that first-order meta-learning algorithms perform well on some well-established benchmarks for few-shot classification, and we provide theoretical analysis aimed at understanding why these algorithms work.

Added

2026-09-14