Dual-Adapter: Training-free Dual Adaptation for Few-shot Out-of-Distribution Detection

TL;DR

This paper introduces Dual-Adapter, a training-free method for few-shot out-of-distribution detection that leverages dual perspectives of features to improve detection accuracy without additional training.

Contribution

The paper proposes a novel training-free dual adaptation approach that utilizes positive and negative feature adapters for effective OOD detection in few-shot scenarios.

Findings

Outperforms existing methods on four benchmark datasets.

Effectively leverages CLIP's capabilities without training.

Improves OOD detection accuracy in few-shot settings.

Abstract

We study the problem of few-shot out-of-distribution (OOD) detection, which aims to detect OOD samples from unseen categories during inference time with only a few labeled in-domain (ID) samples. Existing methods mainly focus on training task-aware prompts for OOD detection. However, training on few-shot data may cause severe overfitting and textual prompts alone may not be enough for effective detection. To tackle these problems, we propose a prior-based Training-free Dual Adaptation method (Dual-Adapter) to detect OOD samples from both textual and visual perspectives. Specifically, Dual-Adapter first extracts the most significant channels as positive features and designates the remaining less relevant channels as negative features. Then, it constructs both a positive adapter and a negative adapter from a dual perspective, thereby better leveraging previously outlooked or interfering features in the training dataset. In this way, Dual-Adapter can inherit the advantages of CLIP not having to train, but also excels in distinguishing between ID and OOD samples. Extensive experimental results on four benchmark datasets demonstrate the superiority of Dual-Adapter.