Dynamic Against Dynamic: An Open-set Self-learning Framework

TL;DR

Open-set recognition often relies on static decision boundaries that fail in dynamic open-world settings. The authors propose Open-Set Self-Learning (OSSL), a dynamic framework that starts from a strong closed-set classifier and adaptively updates a lightweight classifier during testing via a self-matching module that handles known, uncertain, and unknown samples. Key innovations include a three-part self-matching mechanism (C*, D, D'), an enhancement strategy with limited labeled data, and a marginal logit loss to tighten unknown representations. Across standard and cross-dataset benchmarks, OSSL achieves new state-of-the-art results, demonstrating effective use of unknown samples to improve discriminability and robust adaptation to changing distributions.

Abstract

In open-set recognition, existing methods generally learn statically fixed decision boundaries using known classes to reject unknown classes. Though they have achieved promising results, such decision boundaries are evidently insufficient for universal unknown classes in dynamic and open scenarios as they can potentially appear at any position in the feature space. Moreover, these methods just simply reject unknown class samples during testing without any effective utilization for them. In fact, such samples completely can constitute the true instantiated representation of the unknown classes to further enhance the model's performance. To address these issues, this paper proposes a novel dynamic against dynamic idea, i.e., dynamic method against dynamic changing open-set world, where an open-set self-learning (OSSL) framework is correspondingly developed. OSSL starts with a good closed-set classifier trained by known classes and utilizes available test samples for model adaptation during testing, thus gaining the adaptability to changing data distributions. In particular, a novel self-matching module is designed for OSSL, which can achieve the adaptation in automatically identifying known class samples while rejecting unknown class samples which are further utilized to enhance the discriminability of the model as the instantiated representation of unknown classes. Our method establishes new performance milestones respectively in almost all standard and cross-data benchmarks.

Figures (6)

• Figure 1: Differences between existing OSR methods and ours. Existing methods generally learn static and fixed classifiers using known class samples for universal unknown classes, whereas ours follows the dynamic against dynamic idea, i.e., dynamic methods against dynamic changing open-set world, where the learned classifier is dynamic and changeable.
• Figure 2: The overview of our open-set self-learning framework.
• Figure 3: Experiments on the effectiveness of different enhancement strategies.
• Figure 4: Threshold parameter sensitivity analysis on Cifar10 and TinyImageNet, where (a) reports the results for Cifar10 while (b) for TinyImageNet.
• Figure 5: Performance comparisons between our OSSL and two self-supervised learning methods.