Few-shot Novel Category Discovery

TL;DR

This paper introduces Few-Shot Novel Category Discovery (FSNCD), a new setting enabling models to identify known classes and cluster unseen categories with minimal labeled data, improving real-world applicability.

Contribution

It proposes the FSNCD framework and two novel clustering methods, SHC and UKC, to enhance model adaptability in open-set, few-shot learning scenarios.

Findings

Achieves leading performance on five datasets.

Effectively balances known class identification and novel category clustering.

Demonstrates robustness across different task settings.

Abstract

The recently proposed Novel Category Discovery (NCD) adapt paradigm of transductive learning hinders its application in more real-world scenarios. In fact, few labeled data in part of new categories can well alleviate this burden, which coincides with the ease that people can label few of new category data. Therefore, this paper presents a new setting in which a trained agent is able to flexibly switch between the tasks of identifying examples of known (labelled) classes and clustering novel (completely unlabeled) classes as the number of query examples increases by leveraging knowledge learned from only a few (handful) support examples. Drawing inspiration from the discovery of novel categories using prior-based clustering algorithms, we introduce a novel framework that further relaxes its assumptions to the real-world open set level by unifying the concept of model adaptability in few-shot learning. We refer to this setting as Few-Shot Novel Category Discovery (FSNCD) and propose Semi-supervised Hierarchical Clustering (SHC) and Uncertainty-aware K-means Clustering (UKC) to examine the model's reasoning capabilities. Extensive experiments and detailed analysis on five commonly used datasets demonstrate that our methods can achieve leading performance levels across different task settings and scenarios.