Few-Shot Open-Set Recognition using Meta-Learning

TL;DR

This paper introduces PEELER, a meta-learning based algorithm that improves open-set recognition in few-shot and large-scale settings by using randomization, entropy maximization, and Mahalanobis distance, achieving state-of-the-art results.

Contribution

The paper proposes PEELER, a novel meta-learning algorithm that unifies open-set recognition and few-shot classification, addressing overfitting issues of traditional classifiers.

Findings

PEELER outperforms existing methods on CIFAR and miniImageNet.

Achieves significant AUROC improvements for seen/unseen class detection.

Effective in both few-shot and large-scale recognition scenarios.

Abstract

The problem of open-set recognition is considered. While previous approaches only consider this problem in the context of large-scale classifier training, we seek a unified solution for this and the low-shot classification setting. It is argued that the classic softmax classifier is a poor solution for open-set recognition, since it tends to overfit on the training classes. Randomization is then proposed as a solution to this problem. This suggests the use of meta-learning techniques, commonly used for few-shot classification, for the solution of open-set recognition. A new oPen sEt mEta LEaRning (PEELER) algorithm is then introduced. This combines the random selection of a set of novel classes per episode, a loss that maximizes the posterior entropy for examples of those classes, and a new metric learning formulation based on the Mahalanobis distance. Experimental results show that PEELER achieves state of the art open set recognition performance for both few-shot and large-scale recognition. On CIFAR and miniImageNet, it achieves substantial gains in seen/unseen class detection AUROC for a given seen-class classification accuracy.