Learning Unknowns from Unknowns: Diversified Negative Prototypes Generator for Few-Shot Open-Set Recognition

TL;DR

This paper introduces DNPG, a novel method for few-shot open-set recognition that learns from unknown class representations to generate diverse negative prototypes, significantly improving recognition of known and unknown classes with limited data.

Contribution

The paper proposes DNPG, which leverages unknown space learning and a swap alignment module to generate diversified negative prototypes, enhancing FSOR performance.

Findings

Achieves state-of-the-art results on three FSOR datasets.

Effectively models unknown space to improve prototype diversity.

Enhances recognition accuracy for both known and unknown classes.

Abstract

Few-shot open-set recognition (FSOR) is a challenging task that requires a model to recognize known classes and identify unknown classes with limited labeled data. Existing approaches, particularly Negative-Prototype-Based methods, generate negative prototypes based solely on known class data. However, as the unknown space is infinite while the known space is limited, these methods suffer from limited representation capability. To address this limitation, we propose a novel approach, termed \textbf{D}iversified \textbf{N}egative \textbf{P}rototypes \textbf{G}enerator (DNPG), which adopts the principle of "learning unknowns from unknowns." Our method leverages the unknown space information learned from base classes to generate more representative negative prototypes for novel classes. During the pre-training phase, we learn the unknown space representation of the base classes. This representation, along with inter-class relationships, is then utilized in the meta-learning process to construct negative prototypes for novel classes. To prevent prototype collapse and ensure adaptability to varying data compositions, we introduce the Swap Alignment (SA) module. Our DNPG model, by learning from the unknown space, generates negative prototypes that cover a broader unknown space, thereby achieving state-of-the-art performance on three standard FSOR datasets.