MetAdapt: Meta-learned task-adaptive architecture for few-shot classification

Abstract

• We propose a feed-forward model rewiring its architecture according to few-shot task. • We propose meta-learned NAS-cell controllers for feed-forward predictive rewiring. • We show performance preserving way of pruning feed-forward-adaptive NAS-cells. • Using the proposed approach, we obtain strong results on two popular FSC benchmarks. Recently, great progress has been made in the field of Few-Shot Learning (FSL). While many different methods have been proposed, one of the key factors leading to higher FSL performance is surprisingly simple. It is the backbone network architecture used to embed the images of the few-shot tasks. While first works on FSL resorted to small architectures with just a few convolution layers, recent works show that large architectures pre-trained on the training portion of FSL datasets produce strong features that are more easily transferable to novel few-shot tasks, thus attaining significant gains to methods using them. Despite these observations, little to no work has been done towards finding the right backbone for FSL. In this paper we propose MetAdapt that not only meta-searches for an optimized architecture for FSL using Network Architecture Search (NAS), but also results in a model that can adaptively ‘re-wire’ itself predicting the better architecture for a given novel few-shot task. Using the proposed approach we observe strong results on two popular few-shot benchmarks: mini ImageNet and FC100.