Towards Realistic Class-Incremental Learning with Free-Flow Increments

TL;DR

This paper introduces a new realistic setting for class-incremental learning called Free-Flow CIL, and proposes a model-agnostic framework with techniques to improve robustness and stability under variable class arrival streams.

Contribution

The paper formalizes Free-Flow CIL, identifies its challenges, and proposes a robust, model-agnostic framework with specific strategies like CWM and DIWA to enhance performance.

Findings

Existing CIL methods degrade under free-flow class streams.

Proposed strategies improve robustness and stabilize learning.

Experiments show consistent performance gains across baselines.

Abstract

Class-incremental learning (CIL) is typically evaluated under predefined schedules with equal-sized tasks, leaving more realistic and complex cases unexplored. However, a practical CIL system should learns immediately when any number of new classes arrive, without forcing fixed-size tasks. We formalize this setting as Free-Flow Class-Incremental Learning (FFCIL), where data arrives as a more realistic stream with a highly variable number of unseen classes each step. It will make many existing CIL methods brittle and lead to clear performance degradation. We propose a model-agnostic framework for robust CIL learning under free-flow arrivals. It comprises a class-wise mean (CWM) objective that replaces sample frequency weighted loss with uniformly aggregated class-conditional supervision, thereby stabilizing the learning signal across free-flow class increments, as well as method-wise adjustments that improve robustness for representative CIL paradigms. Specifically, we constrain distillation to replayed data, normalize the scale of contrastive and knowledge transfer losses, and introduce Dynamic Intervention Weight Alignment (DIWA) to prevent over-adjustment caused by unstable statistics from small class increments. Experiments confirm a clear performance degradation across various CIL baselines under FFCIL, while our strategies yield consistent gains.