Consistency-Driven Calibration and Matching for Few-Shot Class Incremental Learning

ICLR 2026 Conference Submission15314 Authors

Keywords: Continual learning, Few-shot class-incremental learning

TL;DR: To solve the knowledge conflict problem in FSCIL, we propose Consistency-Driven Calibration and Matching framework.

Abstract: Few-Shot Class Incremental Learning (FSCIL) is crucial for adapting to the complex open-world environments. Contemporary prospective learning-based space construction methods struggle to balance old and new knowledge, as prototype bias and rigid structures limit the expressive capacity of the embedding space. Different from these strategies, we rethink the optimization dilemma from the perspective of feature-structure dual consistency, and propose a Consistency-driven Calibration and Matching (ConCM) framework that systematically mitigates the knowledge conflict inherent in FSCIL. Specifically, inspired by hippocampal associative memory, we design a memory-aware prototype calibration that extracts generalized semantic attributes from base classes and reintegrates them into novel classes to enhance the conceptual center consistency of features. Further, to consolidate memory associations, we propose dynamic structure matching, which adaptively aligns the calibrated features to a session-specific optimal manifold space, ensuring cross-session structure consistency. This process requires no class-number priors and is theoretically guaranteed to achieve geometric optimality and maximum matching. On large-scale FSCIL benchmarks including mini-ImageNet, CIFAR100 and CUB200, ConCM achieves state-of-the-art performance, with harmonic accuracy gains of up to 3.90% in incremental sessions. Code is available at: https://anonymous.4open.science/r/ConCM-7385

Primary Area: transfer learning, meta learning, and lifelong learning

Submission Number: 15314