Finn et al.: initialization optimized for few-step task adaptation

What changed

Optimized shared initial parameters so that a few task-specific gradient updates yield useful performance on a sampled new task, applying the same outer objective across classification, regression and policy learning.

Scientific scope

Optimized shared initial parameters so that a few task-specific gradient updates yield useful performance on a sampled new task, applying the same outer objective across classification, regression and policy learning. Earlier learning-to-learn and gradient optimization are inherited. Model-agnostic concerns compatibility with gradient-trained models, not arbitrary non-differentiable learners or guaranteed out-of-distribution adaptation.

Evidence

The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Earlier learning-to-learn and gradient optimization are inherited. Model-agnostic concerns compatibility with gradient-trained models, not arbitrary non-differentiable learners or guaranteed out-of-distribution adaptation.

Contributors

• Source-listed authors and technical contributors · SCIENCE: Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.

• Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks

Provisional WRS assessment, edition 2026-10-16.1. Independent expert calibration is pending. This grade does not certify correctness or assign ownership.

Full scientific record · Browse contribution grades · Request a correction