When Classes Evolve: A Benchmark and Framework for Stage-Aware Class-Incremental Learning

TL;DR

STAGE is proposed, a novel method that explicitly learns abstract and transferable evolution patterns within a fixed-size memory pool and decoupling semantic identity from transformation dynamics, enabling accurate prediction of future morphologies based on earlier representations.

Abstract

Class-Incremental Learning (CIL) aims to sequentially learn new classes while mitigating catastrophic forgetting of previously learned knowledge. Conventional CIL approaches implicitly assume that classes are morphologically static, focusing primarily on preserving previously learned representations as new classes are introduced. However, this assumption neglects intra-class evolution: a phenomenon wherein instances of the same semantic class undergo significant morphological transformations, such as a larva turning into a butterfly. Consequently, a model must both discriminate between classes and adapt to evolving appearances within a single class. To systematically address this challenge, we formalize Stage-Aware CIL (Stage-CIL), a paradigm in which each class is learned progressively through distinct morphological stages. To facilitate rigorous evaluation within this paradigm, we introduce the Stage-Bench, a 10-domain, 2-stages dataset and protocol that jointly measure inter- and intra-class forgetting. We further propose STAGE, a novel method that explicitly learns abstract and transferable evolution patterns within a fixed-size memory pool. By decoupling semantic identity from transformation dynamics, STAGE enables accurate prediction of future morphologies based on earlier representations. Extensive empirical evaluation demonstrates that STAGE consistently and substantially outperforms existing state-of-the-art approaches, highlighting its effectiveness in simultaneously addressing inter-class discrimination and intra-class morphological adaptation.

Figures (12)

• Figure 1: Conventional CIL focuses on inter-class forgetting across distinct categories. Stage-CIL introduces temporally ordered morphological stages within each class, giving rise to the new problem of intra-class forgetting while retaining the original inter-class challenge.
• Figure 2: Statistical summary of Stage-Bench. It spans 10 diverse domains, each with 20 classes and two ordered stages (Stage-0 and Stage-1), yielding 400 morphological stages in total.
• Figure 3: Illustration of STAGE. Top: the model learns a visual prototype for each new class from initial-morphology images, while text prompts provide class semantics. Bottom: the stored prototype queries the pattern pool, attends to the top-k patterns, and predicts the evolved representation, enabling classification of the Stage 1 images and online pattern updates.
• Figure 4: Incremental performance of different methods. We report the performance gap after the last incremental stage of STAGE and the runner‑up method at the end of the line. All methods are based on the same backbone/weight.
• Figure 5: Ablation study of different components in STAGE. We find each component within STAGE enhances the performance.