One-step adaptation in the intrinsic dimension
Produced by First-order ANIL provably learns representations despite overparametrization
First-order ANIL
A first-order meta-learning algorithm that adapts a task-specific output layer while learning a shared representation across tasks.
A first-order meta-learning algorithm that adapts a task-specific output layer while learning a shared representation across tasks.
It approximates the meta-gradient while restricting task-level adaptation to the final layer.
Connections
Has applicable result
Representation recovery by first-order ANIL
Produced by First-order ANIL provably learns representations despite overparametrization