MOKD: Cross-domain Finetuning for Few-shot Classification via Maximizing Optimized Kernel Dependence

TL;DR

This paper introduces MOKD, a bi-level optimization framework that enhances cross-domain few-shot classification by learning class-specific representations through maximizing kernel dependence, leading to improved generalization and clustering.

Contribution

MOKD is a novel method that optimizes kernel dependence to learn better representations for few-shot classification across domains.

Findings

MOKD outperforms existing methods on Meta-Dataset in most cases.

It learns representations with clearer class clustering.

MOKD improves generalization to unseen domains.

Abstract

In cross-domain few-shot classification, \emph{nearest centroid classifier} (NCC) aims to learn representations to construct a metric space where few-shot classification can be performed by measuring the similarities between samples and the prototype of each class. An intuition behind NCC is that each sample is pulled closer to the class centroid it belongs to while pushed away from those of other classes. However, in this paper, we find that there exist high similarities between NCC-learned representations of two samples from different classes. In order to address this problem, we propose a bi-level optimization framework, \emph{maximizing optimized kernel dependence} (MOKD) to learn a set of class-specific representations that match the cluster structures indicated by labeled data of the given task. Specifically, MOKD first optimizes the kernel adopted in \emph{Hilbert-Schmidt independence criterion} (HSIC) to obtain the optimized kernel HSIC (opt-HSIC) that can capture the dependence more precisely. Then, an optimization problem regarding the opt-HSIC is addressed to simultaneously maximize the dependence between representations and labels and minimize the dependence among all samples. Extensive experiments on Meta-Dataset demonstrate that MOKD can not only achieve better generalization performance on unseen domains in most cases but also learn better data representation clusters. The project repository of MOKD is available at: \href{https://github.com/tmlr-group/MOKD}{https://github.com/tmlr-group/MOKD}.